The Inference Report

Large Chunk Test-Time Training (LaCT) has shown strong performance on long-context 3D reconstruction, but its fully plastic inference-time updates remain vulnerable to catastrophic forgetting and overfitting. As a result, LaCT is typically instantiated with a single large chunk spanning the full input sequence, falling short of the broader goal of handling arbitrarily long sequences in a single pass. We propose Elastic Test-Time Training inspired by elastic weight consolidation, that stabilizes LaCT fast-weight updates with a Fisher-weighted elastic prior around a maintained anchor state. The anchor evolves as an exponential moving average of past fast weights to balance stability and plasticity. Based on this updated architecture, we introduce Fast Spatial Memory (FSM), an efficient and scalable model for 4D reconstruction that learns spatiotemporal representations from long observation sequences and renders novel view-time combinations. We pre-trained FSM on large-scale curated 3D/4D data to capture the dynamics and semantics of complex spatial environments. Extensive experiments show that FSM supports fast adaptation over long sequences and delivers high-quality 3D/4D reconstruction with smaller chunks and mitigating the camera-interpolation shortcut. Overall, we hope to advance LaCT beyond the bounded single-chunk setting toward robust multi-chunk adaptation, a necessary step for generalization to genuinely longer sequences, while substantially alleviating the activation-memory bottleneck.

Exact relevance certification asks which coordinates are necessary to determine the optimal action in a coordinate-structured decision problem. The tractable families treated here admit a finite primitive basis, but optimizer-quotient realizability is maximal, so quotient shape alone cannot characterize the frontier. We prove a meta-impossibility theorem for efficiently checkable structural predicates invariant under the theorem-forced closure laws of exact certification. Structural convergence with zero-distortion summaries, quotient entropy bounds, and support-counting arguments explains why those closure laws are canonical. We establish the theorem by constructing same-orbit disagreements for four obstruction families, namely dominant-pair concentration, margin masking, ghost-action concentration, and additive/statewise offset concentration, using action-independent, pair-targeted affine witnesses. Consequently no correct tractability classifier on a closure-closed domain yields an exact characterization over these families. Here closure-orbit agreement is forced by correctness rather than assumed as an invariance axiom. The result therefore applies to correct classifiers on closure-closed domains, not only to classifiers presented through a designated admissibility package.

Generating motion-controlled videos--where user-specified actions drive physically plausible scene dynamics under freely chosen viewpoints--demands two capabilities: (1) disentangled motion control, allowing users to separately control the object motion and adjust camera viewpoint; and (2) motion causality, ensuring that user-driven actions trigger coherent reactions from other objects rather than merely displacing pixels. Existing methods fall short on both fronts: they entangle camera and object motion into a single tracking signal and treat motion as kinematic displacement without modeling causal relationships between object motion. We introduce MoRight, a unified framework that addresses both limitations through disentangled motion modeling. Object motion is specified in a canonical static-view and transferred to an arbitrary target camera viewpoint via temporal cross-view attention, enabling disentangled camera and object control. We further decompose motion into active (user-driven) and passive (consequence) components, training the model to learn motion causality from data. At inference, users can either supply active motion and MoRight predicts consequences (forward reasoning), or specify desired passive outcomes and MoRight recovers plausible driving actions (inverse reasoning), all while freely adjusting the camera viewpoint. Experiments on three benchmarks demonstrate state-of-the-art performance in generation quality, motion controllability, and interaction awareness.

The rapid growth of generative artificial intelligence (AI) has introduced unprecedented computational demands, driving significant increases in the energy footprint of data centers. However, existing power consumption data is largely proprietary and reported at varying resolutions, creating challenges for estimating whole-facility energy use and planning infrastructure. In this work, we present a methodology that bridges this gap by linking high-resolution workload power measurements to whole-facility energy demand. Using NLR's high-performance computing data center equipped with NVIDIA H100 GPUs, we measure power consumption of AI workloads at 0.1-second resolution for AI training, fine-tuning and inference jobs. Workloads are characterized using MLCommons benchmarks for model training and fine-tuning, and vLLM benchmarks for inference, enabling reproducible and standardized workload profiling. The dataset of power consumption profiles is made publicly available. These power profiles are then scaled to the whole-facility-level using a bottom-up, event-driven, data center energy model. The resulting whole-facility energy profiles capture realistic temporal fluctuations driven by AI workloads and user-behavior, and can be used to inform infrastructure planning for grid connection, on-site energy generation, and distributed microgrids.

Pluralistic alignment has emerged as a critical frontier in the development of Large Language Models (LLMs), with reward models (RMs) serving as a central mechanism for capturing diverse human values. While benchmarks for general response quality are prevalent, evaluating how well reward models account for individual user preferences remains an open challenge. To bridge this gap, we introduce Personalized RewardBench, a novel benchmark designed to rigorously assess reward models' capacity to model personalized preferences. We construct chosen and rejected response pairs based on strict adherence to (or violation of) user-specific rubrics, ensuring that preference distinctions are uniquely tailored to the individual. In particular, human evaluations confirm that the primary discriminative factor between pairs is strictly personal preference, with both responses maintaining high general quality (e.g., correctness, relevance and helpfulness). Extensive testing reveals that existing state-of-the-art reward models struggle significantly with personalization, peaking at an accuracy of just 75.94%. Crucially, because an effective reward model benchmark should predict a reward model's performance on downstream tasks, we conduct experiments demonstrating that our benchmark exhibits a significantly higher correlation with downstream performance in both Best-of-N (BoN) sampling and Proximal Policy Optimization (PPO) compared to existing baselines. These findings establish Personalized RewardBench as a robust and accurate proxy for evaluating reward models' performance in downstream applications.

Most repository-level code translation and validation techniques have been evaluated on a single source-target programming language (PL) pair, owing to the complex engineering effort required to adapt new PL pairs. Programming agents can enable PL-agnosticism in repository-level code translation and validation: they can synthesize code across many PLs and autonomously use existing tools specific to each PL's analysis. However, state-of-the-art has yet to offer a fully autonomous agentic approach for repository-level code translation and validation of large-scale programs. This paper proposes ReCodeAgent, an autonomous multi-agent approach for language-agnostic repository-level code translation and validation. Users only need to provide the project in the source PL and specify the target PL for ReCodeAgent to automatically translate and validate the entire repository. ReCodeAgent is the first technique to achieve high translation success rates across many PLs. We compare the effectiveness of ReCodeAgent with four alternative neuro-symbolic and agentic approaches to translate 118 real-world projects, with 1,975 LoC and 43 translation units for each project, on average. The projects cover 6 PLs (C, Go, Java, JavaScript, Python, and Rust) and 4 PL pairs (C-Rust, Go-Rust, Java-Python, Python-JavaScript). Our results demonstrate that ReCodeAgent consistently outperforms prior techniques on translation correctness, improving test pass rate by 60.8% on ground-truth tests, with an average cost of $15.3. We also perform process-centric analysis of ReCodeAgent trajectories to confirm its procedural efficiency. Finally, we investigate how the design choices (a multi-agent vs. single-agent architecture) influence ReCodeAgent performance: on average, the test pass rate drops by 40.4%, and trajectories become 28% longer and persistently inefficient.

Recent advances in vision-language models (VLMs) have improved image captioning for cultural heritage. However, inferring structured cultural metadata (e.g., creator, origin, period) from visual input remains underexplored. We introduce a multi-category, cross-cultural benchmark for this task and evaluate VLMs using an LLM-as-Judge framework that measures semantic alignment with reference annotations. To assess cultural reasoning, we report exact-match, partial-match, and attribute-level accuracy across cultural regions. Results show that models capture fragmented signals and exhibit substantial performance variation across cultures and metadata types, leading to inconsistent and weakly grounded predictions. These findings highlight the limitations of current VLMs in structured cultural metadata inference beyond visual perception.

Scaling up robot learning will likely require human data containing rich and long-horizon interactions in the wild. Existing approaches for collecting such data trade off portability, robustness to occlusion, and global consistency. We introduce RoSHI, a hybrid wearable that fuses low-cost sparse IMUs with the Project Aria glasses to estimate the full 3D pose and body shape of the wearer in a metric global coordinate frame from egocentric perception. This system is motivated by the complementarity of the two sensors: IMUs provide robustness to occlusions and high-speed motions, while egocentric SLAM anchors long-horizon motion and stabilizes upper body pose. We collect a dataset of agile activities to evaluate RoSHI. On this dataset, we generally outperform other egocentric baselines and perform comparably to a state-of-the-art exocentric baseline (SAM3D). Finally, we demonstrate that the motion data recorded from our system are suitable for real-world humanoid policy learning. For videos, data and more, visit the project webpage: https://roshi-mocap.github.io/

How does the choice of training data influence an AI model? This question is of central importance to interpretability, privacy, and basic science. At its core is the data deletion problem: after a reasonable amount of precomputation, quickly predict how the model would behave in a given situation if a given subset of training data had been excluded from the learning algorithm. We present a data deletion scheme capable of predicting model outputs with vanishing error $\varepsilon$ in the deep learning setting. Our precomputation and prediction algorithms are only $\mathrm{poly}(1/\varepsilon)$ factors slower than regular training and inference, respectively. The storage requirements are those of $\mathrm{poly}(1/\varepsilon)$ models. Our proof is based on an assumption that we call "stability." In contrast to the assumptions made by prior work, stability appears to be fully compatible with learning powerful AI models. In support of this, we show that stability is satisfied in a minimal set of experiments with microgpt. Our code is available at https://github.com/SamSpo1/microgpt-sketch. At a technical level, our work is based on a new method for locally sketching an arithmetic circuit by computing higher-order derivatives in random complex directions. Forward-mode automatic differentiation allows cheap computation of these derivatives.

In this paper, we derive rates of convergence in the high-dimensional central limit theorem for Polyak-Ruppert averaged iterates generated by the asynchronous Q-learning algorithm with a polynomial stepsize $k^{-ω},\, ω\in (1/2, 1]$. Assuming that the sequence of state-action-next-state triples $(s_k, a_k, s_{k+1})_{k \geq 0}$ forms a uniformly geometrically ergodic Markov chain, we establish a rate of order up to $n^{-1/6} \log^{4} (nS A)$ over the class of hyper-rectangles, where $n$ is the number of samples used by the algorithm and $S$ and $A$ denote the numbers of states and actions, respectively. To obtain this result, we prove a high-dimensional central limit theorem for sums of martingale differences, which may be of independent interest. Finally, we present bounds for high-order moments for the algorithm's last iterate.

Propositional Linear Temporal Logic (LTL) is a popular formalism for specifying desirable requirements and security and privacy policies for software, networks, and systems. Yet expressing such requirements and policies in LTL remains challenging because of its intricate semantics. Since many security and privacy analysis tools require LTL formulas as input, this difficulty places them out of reach for many developers and analysts. Large Language Models (LLMs) could broaden access to such tools by translating natural language fragments into LTL formulas. This paper evaluates that premise by assessing how effectively several representative LLMs translate assertive English sentences into LTL formulas. Using both human-generated and synthetic ground-truth data, we evaluate effectiveness along syntactic and semantic dimensions. The results reveal three findings: (1) in line with prior findings, LLMs perform better on syntactic aspects of LTL than on semantic ones; (2) they generally benefit from more detailed prompts; and (3) reformulating the task as a Python code-completion problem substantially improves overall performance. We also discuss challenges in conducting a fair evaluation on this task and conclude with recommendations for future work.

Low-resource languages pose a challenge for machine translation with large language models (LLMs), which require large amounts of training data. One potential way to circumvent this data dependence is to rely on LLMs' ability to use in-context descriptions of languages, like textbooks and dictionaries. To do so, LLMs must be able to infer the link between the languages' grammatical descriptions and the sentences in question. Here we isolate this skill using a formal analogue of the task: string transduction based on a formal grammar provided in-context. We construct synchronous context-free grammars which define pairs of formal languages designed to model particular aspects of natural language grammar, morphology, and written representation. Using these grammars, we measure how well LLMs can translate sentences from one formal language into another when given both the grammar and the source-language sentence. We vary the size of the grammar, the lengths of the sentences, the syntactic and morphological properties of the languages, and their written script. We note three key findings. First, LLMs' translation accuracy decreases markedly as a function of grammar size and sentence length. Second, differences in morphology and written representation between the source and target languages can strongly diminish model performance. Third, we examine the types of errors committed by models and find they are most prone to recall the wrong words from the target language vocabulary, hallucinate new words, or leave source-language words untranslated.

The growing complexity of neural networks hinders the deployment of distributed machine learning on resource-constrained devices. Split learning (SL) offers a promising solution by partitioning the large model and offloading the primary training workload from edge devices to an edge server. However, the increasing number of participating devices and model complexity leads to significant communication overhead from the transmission of smashed data (e.g., activations and gradients), which constitutes a critical bottleneck for SL. To tackle this challenge, we propose SL-FAC, a communication-efficient SL framework comprising two key components: adaptive frequency decomposition (AFD) and frequency-based quantization compression (FQC). AFD first transforms the smashed data into the frequency domain and decomposes it into spectral components with distinct information. FQC then applies customized quantization bit widths to each component based on its spectral energy distribution. This collaborative approach enables SL-FAC to achieve significant communication reduction while strategically preserving the information most crucial for model convergence. Extensive experiments confirm the superior performance of SL-FAC for improving the training efficiency.

Existing dynamic data pruning methods often fail under noisy-label settings, as they typically rely on per-sample loss as the ranking criterion. This could mistakenly lead to preserving noisy samples due to their high loss values, resulting in significant performance drop. To address this, we propose AlignPrune, a noise-robust module designed to enhance the reliability of dynamic pruning under label noise. Specifically, AlignPrune introduces the Dynamic Alignment Score (DAS), which is a loss-trajectory-based criterion that enables more accurate identification of noisy samples, thereby improving pruning effectiveness. As a simple yet effective plug-and-play module, AlignPrune can be seamlessly integrated into state-of-the-art dynamic pruning frameworks, consistently outperforming them without modifying either the model architecture or the training pipeline. Extensive experiments on five widely-used benchmarks across various noise types and pruning ratios demonstrate the effectiveness of AlignPrune, boosting accuracy by up to 6.3\% over state-of-the-art baselines. Our results offer a generalizable solution for pruning under noisy data, encouraging further exploration of learning in real-world scenarios. Code is available at: https://github.com/leonqin430/AlignPrune.

Large Language Models (LLMs) challenge conventional automated programming assessment because students can now produce functionally correct code without demonstrating corresponding understanding. This paper makes two contributions. First, it reports a saturation-based scoping review of conversational assessment approaches in programming education. The review identifies three dominant architectural families: rule-based or template-driven systems, LLM-based systems, and hybrid systems. Across the literature, conversational agents appear promising for scalable feedback and deeper probing of code understanding, but important limitations remain around hallucinations, over-reliance, privacy, integrity, and deployment constraints. Second, the paper synthesizes these findings into a Hybrid Socratic Framework for integrating conversational verification into Automated Programming Assessment Systems (APASs). The framework combines deterministic code analysis with a dual-agent conversational layer, knowledge tracking, scaffolded questioning, and guardrails that tie prompts to runtime facts. The paper also discusses practical safeguards against LLM-generated explanations, including proctored deployment modes, randomized trace questions, stepwise reasoning tied to concrete execution states, and local-model deployment options for privacy-sensitive settings. Rather than replacing conventional testing, the framework is intended as a complementary layer for verifying whether students understand the code they submit.

Multiple Instance Learning (MIL) is the dominant framework for gigapixel whole-slide image (WSI) classification in computational pathology. However, current MIL aggregators route all instances through a shared pathway, constraining their capacity to specialise across the pathological heterogeneity inherent in each slide. Mixture-of-Experts (MoE) methods offer a natural remedy by partitioning instances across specialised expert subnetworks; yet unconstrained softmax routing may yield highly imbalanced utilisation, where one or a few experts absorb most routing mass, collapsing the mixture back to a near-single-pathway solution. To address these limitations, we propose ROAM (Region-graph OptimAl-transport Mixture-of-experts), a spatially aware MoE-MIL aggregator that routes region tokens to expert poolers via capacity-constrained entropic optimal transport, promoting balanced expert utilisation by construction. ROAM operates on spatial region tokens, obtained by compressing dense patch bags into spatially binned units that align routing with local tissue neighbourhoods and introduces two key mechanisms: (i) region-to-expert assignment formulated as entropic optimal transport (Sinkhorn) with explicit per slide capacity marginals, enforcing balanced expert utilisation without auxiliary load-balancing losses; and (ii) graph-regularised Sinkhorn iterations that diffuse routing assignments over the spatial region graph, encouraging neighbouring regions to coherently route to the same experts. Evaluated on four WSI benchmarks with frozen foundation-model patch embeddings, ROAM achieves performance competitive against strong MIL and MoE baselines, and on NSCLC generalisation (TCGA-CPTAC) reaches external AUC 0.845 +- 0.019.

Spatial understanding is a fundamental cornerstone of human-level intelligence. Nonetheless, current research predominantly focuses on domain-specific data production, leaving a critical void: the absence of a principled, open-source engine capable of fully unleashing the potential of high-quality spatial data. To bridge this gap, we elucidate the design principles of a robust data generation system and introduce OpenSpatial -- an open-source data engine engineered for high quality, extensive scalability, broad task diversity, and optimized efficiency. OpenSpatial adopts 3D bounding boxes as the fundamental primitive to construct a comprehensive data hierarchy across five foundational tasks: Spatial Measurement (SM), Spatial Relationship (SR), Camera Perception (CP), Multi-view Consistency (MC), and Scene-Aware Reasoning (SAR). Leveraging this scalable infrastructure, we curate OpenSpatial-3M, a large-scale dataset comprising 3 million high-fidelity samples. Extensive evaluations demonstrate that versatile models trained on our dataset achieve state-of-the-art performance across a wide spectrum of spatial reasoning benchmarks. Notably, the best-performing model exhibits a substantial average improvement of 19 percent, relatively. Furthermore, we provide a systematic analysis of how data attributes influence spatial perception. By open-sourcing both the engine and the 3M-scale dataset, we provide a robust foundation to accelerate future research in spatial intelligence.

Real-time supervisory control of advanced reactors requires accurate forecasting of plant-wide thermal-hydraulic states, including locations where physical sensors are unavailable. Meeting this need calls for surrogate models that combine predictive fidelity, millisecond-scale inference, and robustness to partial observability. In this work, we present a physics-informed message-passing Graph Neural Network coupled with a Neural Ordinary Differential Equation (GNN-ODE) to addresses all three requirements simultaneously. We represent the whole system as a directed sensor graph whose edges encode hydraulic connectivity through flow/heat transfer-aware message passing, and we advance the latent dynamics in continuous time via a controlled Neural ODE. A topology-guided missing-node initializer reconstructs uninstrumented states at rollout start; prediction then proceeds fully autoregressively. The GNN-ODE surrogate achieves satisfactory results for the system dynamics prediction. On held-out simulation transients, the surrogate achieves an average MAE of 0.91 K at 60 s and 2.18 K at 300 s for uninstrumented nodes, with $R^2$ up to 0.995 for missing-node state reconstruction. Inference runs at approximately 105 times faster than simulated time on a single GPU, enabling 64-member ensemble rollouts for uncertainty quantification. To assess sim-to-real transfer, we adapt the pretrained surrogate to experimental facility data using layerwise discriminative fine-tuning with only 30 training sequences. The learned flow-dependent heat-transfer scaling recovers a Reynolds-number exponent consistent with established correlations, indicating constitutive learning beyond trajectory fitting. The model tracks a steep power change transient and produces accurate trajectories at uninstrumented locations.

Autonomous vehicles deployed in remote environments typically rely on embedded processors, compact batteries, and lightweight sensors. These hardware limitations conflict with the need to derive robust representations of the environment, which often requires executing computationally intensive deep neural networks for perception. To address this challenge, we present CADENCE, an adaptive system that dynamically scales the computational complexity of a slimmable monocular depth estimation network in response to navigation needs and environmental context. By closing the loop between perception fidelity and actuation requirements, CADENCE ensures high-precision computing is only used when mission-critical. We conduct evaluations on our released open-source testbed that integrates Microsoft AirSim with an NVIDIA Jetson Orin Nano. As compared to a state-of-the-art static approach, CADENCE decreases sensor acquisitions, power consumption, and inference latency by 9.67%, 16.1%, and 74.8%, respectively. The results demonstrate an overall reduction in energy expenditure by 75.0%, along with an increase in navigation accuracy by 7.43%.

Debates about artificial intelligence (AI) in education often portray teaching as a modular and procedural job that can increasingly be automated or delegated to technology. This brief communication paper argues that such claims depend on treating teaching as more separable than it is in practice. Drawing on recent literature and empirical studies of large language models and retrieval-augmented generation systems, I argue that although AI can support some bounded functions, instructional work remains difficult to automate in meaningful ways because it is inherently interpretive, relational, and grounded in professional judgment. More fundamentally, teaching and learning are shaped by human cognition, behavior, motivation, and social interaction in ways that cannot be fully specified, predicted, or exhaustively modeled. Tasks that may appear separable in principle derive their instructional value in practice from ongoing contextual interpretation across learners, situations, and relationships. As long as educational practice relies on emergent understanding of human cognition and learning, teaching remains a form of professional work that resists automation. AI may improve access to information and support selected instructional activities, but it does not remove the need for human judgment and relational accountability that effective teaching requires.

Automated face recognition has made rapid strides over the past decade due to the unprecedented rise of deep neural network (DNN) models that can be trained for domain-specific tasks. At the same time, foundation models that are pretrained on broad vision or vision-language tasks have shown impressive generalization across diverse domains, including biometrics. This raises an important question: Do different DNN models--both domain-specific and foundation models--encode facial identity in similar ways, despite being trained on different datasets, loss functions, and architectures? In this regard, we directly analyze the geometric structure of embedding spaces imputed by different DNN models. Treating embeddings of face images as point clouds, we study whether simple affine transformations can align face representations of one model with another. Our findings reveal surprising cross-model compatibility: low-capacity linear mappings substantially improve cross-model face recognition over unaligned baselines for both face identification and verification tasks. Alignment patterns generalize across datasets and vary systematically across model families, indicating representational convergence in facial identity encoding. These findings have implications for model interoperability, ensemble design, and biometric template security.

Online reinforcement learning (RL) serves as an effective method for enhancing the capabilities of Android agents. However, guiding agents to learn through online interaction is prohibitively expensive due to the high latency of emulators and the sample inefficiency of existing RL algorithms. We identify a fundamental limitation in current approaches: the Single State Single Action paradigm, which updates the policy with one-to-one state-action pairs from online one-way rollouts without fully exploring each costly emulator state. In this paper, we propose Android Coach, a novel framework that shifts the training paradigm to Single State Multiple Actions, allowing the agent to sample and utilize multiple actions for a single online state. We enable this without additional emulator overhead by learning a critic that estimates action values. To ensure the critic serves as a reliable coach, we integrate a process reward model and introduce a group-wise advantage estimator based on the averaged critic outputs. Extensive experiments demonstrate the effectiveness and efficiency of Android Coach: it achieves 7.5% and 8.3% success rate improvements on AndroidLab and AndroidWorld over UI-TARS-1.5-7B, and attains 1.4x higher training efficiency than Single State Single Action methods PPO and GRPO at matched success rates.

GROMACS is a de-facto standard for classical Molecular Dynamics (MD). The rise of AI-driven interatomic potentials that pursue near-quantum accuracy at MD throughput now poses a significant challenge: embedding neural-network inference into multi-GPU simulations retaining high-performance. In this work, we integrate the MLIP framework DeePMD-kit into GROMACS, enabling domain-decomposed, GPU-accelerated inference across multi-node systems. We extend the GROMACS NNPot interface with a DeePMD backend, and we introduce a domain decomposition layer decoupled from the main simulation. The inference is executed concurrently on all processes, with two MPI collectives used each step to broadcast coordinates and to aggregate and redistribute forces. We train an in-house DPA-1 model (1.6 M parameters) on a dataset of solvated protein fragments. We validate the implementation on a small protein system, then we benchmark the GROMACS-DeePMD integration with a 15,668 atom protein on NVIDIA A100 and AMD MI250x GPUs up to 32 devices. Strong-scaling efficiency reaches 66% at 16 devices and 40% at 32; weak-scaling efficiency is 80% to 16 devices and reaches 48% (MI250x) and 40% (A100) at 32 devices. Profiling with the ROCm System profiler shows that >90% of the wall time is spent in DeePMD inference, while MPI collectives contribute <10%, primarily since they act as a global synchronization point. The principal bottlenecks are the irreducible ghost-atom cost set by the cutoff radius, confirmed by a simple throughput model, and load imbalance across ranks. These results demonstrate that production MD with near ab initio fidelity is feasible at scale in GROMACS.

Large language models (LLMs) have demonstrated strong capabilities in medical question answering; however, purely parametric models often suffer from knowledge gaps and limited factual grounding. Retrieval-augmented generation (RAG) addresses this limitation by integrating external knowledge retrieval into the reasoning process. Despite increasing interest in RAG-based medical systems, the impact of individual retrieval components on performance remains insufficiently understood. This study presents a systematic evaluation of retrieval-augmented medical question answering using the MedQA USMLE benchmark and a structured textbook-based knowledge corpus. We analyze the interaction between language models, embedding models, retrieval strategies, query reformulation, and cross-encoder reranking within a unified experimental framework comprising forty configurations. Results show that retrieval augmentation significantly improves zero-shot medical question answering performance. The best-performing configuration was dense retrieval with query reformulation and reranking achieved 60.49% accuracy. Domain-specialized language models were also found to better utilize retrieved medical evidence than general-purpose models. The analysis further reveals a clear tradeoff between retrieval effectiveness and computational cost, with simpler dense retrieval configurations providing strong performance while maintaining higher throughput. All experiments were conducted on a single consumer-grade GPU, demonstrating that systematic evaluation of retrieval-augmented medical QA systems can be performed under modest computational resources.

The widespread use of clickbait headlines, crafted to mislead and maximize engagement, poses a significant challenge to online credibility. These headlines employ sensationalism, misleading claims, and vague language, underscoring the need for effective detection to ensure trustworthy digital content. The paper introduces, ClickGuard: a trustworthy adaptive fusion framework for clickbait detection. It combines BERT embeddings and structural features using a Syntactic-Semantic Adaptive Fusion Block (SSAFB) for dynamic integration. The framework incorporates a hybrid CNN-BiLSTM to capture patterns and dependencies. The model achieved 96.93% testing accuracy, outperforming state-of-the-art approaches. The model's trustworthiness is evaluated using LIME and Permutation Feature Importance (PFI) for interpretability and perturbation analysis. These methods assess the model's robustness and sensitivity to feature changes by measuring the average prediction variation. Ablation studies validated the SSAFB's effectiveness in optimizing feature fusion. The model demonstrated robust performance across diverse datasets, providing a scalable, reliable solution for enhancing online content credibility by addressing syntactic-semantic modelling challenges. Code of the work is available at: https://github.com/palindromeRice/ClickBait_Detection_Architecture

Clinical expertise improves not only by acquiring medical knowledge, but by accumulating experience that yields reusable diagnostic patterns. Recent LLMs-based diagnostic agents have shown promising progress in clinical reasoning for decision support. However, most approaches treat cases independently, limiting experience reuse and continual adaptation. We propose SEA, a self-learning diagnostic agent with cognitively inspired dual-memory module. We design a reinforcement training framework tailored to our designed agent for joint optimization of reasoning and memory management. We evaluate SEA in two complementary settings. On standard evaluation with MedCaseReasoning dataset, SEA achieves 92.46% accuracy, outperforming the strongest baseline by +19.6%, demonstrating the benefit of jointly optimizing reasoning and memory. On the long-horizon with ER-Reason dataset, SEA attains the best final accuracy (0.7214) and the largest improvement (+0.35 Acc@100), while baseline methods show limited or unstable gains. Expert evaluation further indicates that rules consolidated from SEA show strong clinical correctness, usefulness and trust, suggesting that the induced rules in dual-memory module are reliable and practically meaningful. Overall, SEA improves both diagnostic reasoning ability and continual learning by effectively transforming experience into reusable knowledge.

Gaussian process ($GP$) regression is a widely used non-parametric modeling tool, but its cubic complexity in the training size limits its use on massive data sets. A practical remedy is to predict using only the nearest neighbours of each test point, as in Nearest Neighbour Gaussian Process ($NNGP$) regression for geospatial problems and the related scalable $GPnn$ method for more general machine-learning applications. Despite their strong empirical performance, the large-$n$ theory of $NNGP/GPnn$ remains incomplete. We develop a theoretical framework for $NNGP$ and $GPnn$ regression. Under mild regularity assumptions, we derive almost sure pointwise limits for three key predictive criteria: mean squared error ($MSE$), calibration coefficient ($CAL$), and negative log-likelihood ($NLL$). We then study the $L_2$-risk, prove universal consistency, and show that the risk attains Stone's minimax rate $n^{-2α/(2p+d)}$, where $α$ and $p$ capture regularity of the regression problem. We also prove uniform convergence of $MSE$ over compact hyper-parameter sets and show that its derivatives with respect to lengthscale, kernel scale, and noise variance vanish asymptotically, with explicit rates. This explains the observed robustness of $GPnn$ to hyper-parameter tuning. These results provide a rigorous statistical foundation for $NNGP/GPnn$ as a highly scalable and principled alternative to full $GP$ models.

Evaluating robustness under temporal distribution shift remains an open challenge. Existing metrics quantify the average decline in performance, but fail to capture how models adapt to evolving data. As a result, temporal degradation is often misinterpreted: when accuracy declines, it is unclear whether the model is failing to adapt or whether the data itself has become inherently more challenging to learn. In this work, we propose three complementary metrics to distinguish adaptation from intrinsic difficulty in the data. Together, these metrics provide a dynamic and interpretable view of model behavior under temporal distribution shift. Results show that our metrics uncover adaptation patterns hidden by existing analysis, offering a richer understanding of temporal robustness in evolving environments.

Operating LEO mega-constellations requires translating high-level operator intents ("reroute financial traffic away from polar links under 80 ms") into low-level routing constraints -- a task that demands both natural language understanding and network-domain expertise. We present an end-to-end system comprising three components: (1) a GNN cost-to-go router that distills Dijkstra-quality routing into a 152K-parameter graph attention network achieving 99.8% packet delivery ratio with 17x inference speedup; (2) an LLM intent compiler that converts natural language to a typed constraint intermediate representation using few-shot prompting with a verifier-feedback repair loop, achieving 98.4% compilation rate and 87.6% full semantic match on feasible intents in a 240-intent benchmark (193 feasible, 47 infeasible); and (3) an 8-pass deterministic validator with constructive feasibility certification that achieves 0% unsafe acceptance on all 47 infeasible intents (30 labeled + 17 discovered by Pass 8), with 100% corruption detection across 240 structural corruption tests and 100% on 15 targeted adversarial attacks. End-to-end evaluation across four constrained routing scenarios confirms zero constraint violations with both routers. We further demonstrate that apparent performance gaps in polar-avoidance scenarios are largely explained by topological reachability ceilings rather than routing quality, and that the LLM compiler outperforms a rule-based baseline by 46.2 percentage points on compositional intents. Our system bridges the semantic gap between operator intent and network configuration while maintaining the safety guarantees required for operational deployment.

In this growing age of data and technology, large black-box models are becoming the norm due to their ability to handle vast amounts of data and learn incredibly complex data patterns. The deficiency of these methods, however, is their inability to explain the prediction process, making them untrustworthy and their use precarious in high-stakes situations. SHapley Additive exPlanations (SHAP) analysis is an explainable AI method growing in popularity for its ability to explain model predictions in terms of the original features. For each sample and feature in the data set, an associated SHAP value quantifies the contribution of that feature to the prediction of that sample. Analysis of these SHAP values provides valuable insight into the model's decision-making process, which can be leveraged to create data-driven solutions. The interpretation of these SHAP values, however, is model-dependent, so there does not exist a universal analysis procedure. To aid in these efforts, we present a detailed investigation of SHAP analysis across various machine learning models and data sets. In uncovering the details and nuance behind SHAP analysis, we hope to empower analysts in this less-explored territory. We also present a novel generalization of the waterfall plot to the multi-classification problem.

Deep neural networks can predict human judgments, but this does not imply that they rely on human-like information or reveal the cues underlying those judgments. Prior work has addressed this issue using attribution heatmaps, but their explanatory value in itself depends on robustness. Here we tested the robustness of such explanations by evaluating whether models that predict human authenticity ratings also produce consistent explanations within and across architectures. We fit lightweight regression heads to multiple frozen pretrained vision models and generated attribution maps using Grad-CAM, LIME, and multiscale pixel masking. Several architectures predicted ratings well, reaching about 80% of the noise ceiling. VGG models achieved this by tracking image quality rather than authenticity-specific variance, limiting the relevance of their attributions. Among the remaining models, attribution maps were generally stable across random seeds within an architecture, especially for EfficientNetB3 and Barlow Twins, and consistency was higher for images judged as more authentic. Crucially, agreement in attribution across architectures was weak even when predictive performance was similar. To address this, we combined models in ensembles, which improved prediction of human authenticity judgments and enabled image-level attribution via pixel masking. We conclude that while deep networks can predict human authenticity judgments well, they do not produce identifiable explanations for those judgments. More broadly, our findings suggest that post hoc explanations from successful models of behavior should be treated as weak evidence for cognitive mechanism.

In gender-restrictive and surveilled contexts, where access to formal education may be restricted for women, pursuing education involves safety and privacy risks. When women are excluded from schools and universities, they often turn to online self-learning and generative AI (GenAI) to pursue their educational and career aspirations. However, we know little about what safe and accountable GenAI support is required in the context of surveillance, household responsibilities, and the absence of learning communities. We present a remote participatory design study with 20 women in Afghanistan, informed by a recruitment survey (n = 140), examining how participants envision GenAI for learning and employability. Participants describe using GenAI less as an information source and more as an always-available peer, mentor, and source of career guidance that helps compensate for the absence of learning communities. At the same time, they emphasize that this companionship is constrained by privacy and surveillance risks, contextually unrealistic and culturally unsafe support, and direct-answer interactions that can undermine learning by creating an illusion of progress. Beyond eliciting requirements, envisioning the future with GenAI through participatory design was positively associated with significant increases in participants' aspirations (p=.01), perceived agency (p=.01), and perceived avenues (p=.03). These outcomes show that accountable and safe GenAI is not only about harm reduction but can also actively enable women to imagine and pursue viable learning and employment futures. Building on this, we translate participants' proposals into accountability-focused design directions that center on safety-first interaction and user control, context-grounded support under constrained resources, and offer pedagogically aligned assistance that supports genuine learning rather than quick answers.

Despite deep learning models running well-defined mathematical functions, we lack a formal mathematical framework for describing model architectures. Ad-hoc notation, diagrams, and pseudocode poorly handle nonlinear broadcasting and the relationship between individual components and composed models. This paper introduces a categorical framework for deep learning models that formalizes broadcasting through the novel axis-stride and array-broadcasted categories. This allows the mathematical function underlying architectures to be precisely expressed and manipulated in a compositional manner. These mathematical definitions are translated into human manageable diagrams and machine manageable data structures. We provide a mirrored implementation in Python (pyncd) and TypeScript (tsncd) to show the universal aspect of our framework, along with features including algebraic construction, graph conversion, PyTorch compilation and diagram rendering. This lays the foundation for a systematic, formal approach to deep learning model design and analysis.

We introduce a code-based challenge for automated, open-ended mathematical discovery based on the $k$-server conjecture, a central open problem in competitive analysis. The task is to discover a potential function satisfying a large graph-structured system of simple linear inequalities. The resulting evaluation procedure is sound but incomplete: any violated inequality definitively refutes a candidate, whereas satisfying all inequalities does not by itself constitute a proof of the corresponding conjecture's special case. Nevertheless, a candidate that passes all constraints would be strong evidence toward a valid proof and, to the best of our knowledge, no currently known potential achieves this under our formulation in the open $k=4$ circle case. As such, a successful candidate would already be an interesting contribution to the $k$-server conjecture, and could become a substantial theoretical result when paired with a full proof. Experiments on the resolved $k=3$ regime show that current agentic methods can solve nontrivial instances, and in the open $k=4$ regime they reduce the number of violations relative to existing potentials without fully resolving the task. Taken together, these results suggest that the task is challenging but plausibly within reach of current methods. Beyond its relevance to the $k$-server community, where the developed tooling enables researchers to test new hypotheses and potentially improve on the current record, the task also serves as a useful \emph{benchmark} for developing code-based discovery agents. In particular, our $k=3$ results show that it mitigates important limitations of existing open-ended code-based benchmarks, including early saturation and the weak separation between naive random baselines and more sophisticated methods.

While Learned Data Compression (LDC) has achieved superior compression ratios, balancing precise probability modeling with system efficiency remains challenging. Crucially, uniform single-stream architectures struggle to simultaneously capture micro-syntactic and macro-semantic features, necessitating deep serial stacking that exacerbates latency. Compounding this, heterogeneous systems are constrained by device speed mismatches, where throughput is capped by Amdahl's Law due to serial processing. To this end, we propose a Dual-Stream Multi-Scale Decoupler that disentangles local and global contexts to replace deep serial processing with shallow parallel streams, and incorporate a Hierarchical Gated Refiner for adaptive feature refinement and precise probability modeling. Furthermore, we design a Concurrent Stream-Parallel Pipeline, which overcomes systemic bottlenecks to achieve full-pipeline parallelism. Extensive experiments demonstrate that our method achieves state-of-the-art performance in both compression ratio and throughput, while maintaining the lowest latency and memory usage. The code is available at https://github.com/huidong-ma/FADE.

As large language models (LLMs) are increasingly trained on sensitive user data, understanding the fundamental cost of privacy in language learning becomes essential. We initiate the study of differentially private (DP) language identification and generation in the agnostic statistical setting, establishing algorithms and matching lower bounds that precisely quantify the cost of privacy. For both tasks, approximate $(\varepsilon, δ)$-DP with constant $\varepsilon > 0$ recovers the non-private error rates: $\exp(-r(n))$ for identification (for any $r(n) = o(n)$) and $\exp(-Ω(n))$ for generation. Under pure $\varepsilon$-DP, the exponents degrade by a multiplicative factor of $\min\{1, \varepsilon\}$, which we show is tight up to constants. Notably, for generation under pure DP with mild assumptions, the upper bound $\exp(-\min\{1,\varepsilon\} \cdot Ω(n))$ matches the lower bound up to some constants, establishing an optimal rate. Our results show that the cost of privacy in language learning is surprisingly mild: absent entirely under approximate DP, and exactly a $\min\{1,\varepsilon\}$ factor in the exponent under pure DP.

Recent LLM-based agents often place world modeling, planning, and reflection inside a single language model loop. This can produce capable behavior, but it makes a basic scientific question difficult to answer: which part of the agent's competence actually comes from the LLM, and which part comes from explicit structure around it? We study this question not by claiming a general answer, but by making it empirically tractable. We introduce a declared reflective runtime protocol that externalizes agent state, confidence signals, guarded actions, and hypothetical transitions into inspectable runtime structure. We instantiate this protocol in a declarative runtime and evaluate it on noisy Collaborative Battleship [4] using four progressively structured agents over 54 games (18 boards $\times$ 3 seeds). The resulting decomposition isolates four components: posterior belief tracking, explicit world-model planning, symbolic in-episode reflection, and sparse LLM-based revision. Across this decomposition, explicit world-model planning improves substantially over a greedy posterior-following baseline (+24.1pp win rate, +0.017 F1). Symbolic reflection operates as a real runtime mechanism -- with prediction tracking, confidence gating, and guarded revision actions -- even though its current revision presets are not yet net-positive in aggregate. Adding conditional LLM revision at about 4.3\% of turns yields only a small and non-monotonic change: average F1 rises slightly (+0.005) while win rate drops (31$\rightarrow$29 out of 54). These results suggest a methodological contribution rather than a leaderboard claim: externalizing reflection turns otherwise latent agent behavior into inspectable runtime structure, allowing the marginal role of LLM intervention to be studied directly.

We provide a computational complexity lens to understand the power of machine learning models, particularly their ability to model complex systems. Machine learning models are often trained on data drawn from sampleable or more complex distributions, a far wider range of distributions than just computable ones. By focusing on computable distributions, machine learning models can better manage complexity via probability. We abstract away from specific learning mechanisms, modeling machine learning as producing P/poly-computable distributions with polynomially-bounded max-entropy. We illustrate how learning computable distributions models complexity by showing that if a machine learning model produces a distribution $μ$ that minimizes error against the distribution generated by a cryptographic pseudorandom generator, then $μ$ must be close to uniform.

As large language models (LLMs) evolve from static chatbots into autonomous agents, the primary vulnerability surface shifts from final outputs to intermediate execution traces. While safety guardrails are well-benchmarked for natural language responses, their efficacy remains largely unexplored within multi-step tool-use trajectories. To address this gap, we introduce TraceSafe-Bench, the first comprehensive benchmark specifically designed to assess mid-trajectory safety. It encompasses 12 risk categories, ranging from security threats (e.g., prompt injection, privacy leaks) to operational failures (e.g., hallucinations, interface inconsistencies), featuring over 1,000 unique execution instances. Our evaluation of 13 LLM-as-a-guard models and 7 specialized guardrails yields three critical findings: 1) Structural Bottleneck: Guardrail efficacy is driven more by structural data competence (e.g., JSON parsing) than semantic safety alignment. Performance correlates strongly with structured-to-text benchmarks ($ρ=0.79$) but shows near-zero correlation with standard jailbreak robustness. 2) Architecture over Scale: Model architecture influences risk detection performance more significantly than model size, with general-purpose LLMs consistently outperforming specialized safety guardrails in trajectory analysis. 3) Temporal Stability: Accuracy remains resilient across extended trajectories. Increased execution steps allow models to pivot from static tool definitions to dynamic execution behaviors, actually improving risk detection performance in later stages. Our findings suggest that securing agentic workflows requires jointly optimizing for structural reasoning and safety alignment to effectively mitigate mid-trajectory risks.

High-dimensional data are often modeled as lying near a low-dimensional manifold. We study how to construct diffusion processes on this data manifold in the implicit setting. That is, using only point cloud samples and without access to charts, projections, or other geometric primitives. Our main contribution is a data-driven SDE that captures intrinsic diffusion on the underlying manifold while being defined in ambient space. The construction relies on estimating the diffusion's infinitesimal generator and its carré-du-champ (CDC) from a proximity graph built from the data. The generator and CDC together encode the local stochastic and geometric structure of the intended diffusion. We show that, as the number of samples grows, the induced process converges in law on the space of probability paths to its smooth manifold counterpart. We call this construction Implicit Manifold-valued Diffusions (IMDs), and furthermore present a numerical simulation procedure using Euler-Maruyama integration. This gives a rigorous basis for practical implementations of diffusion dynamics on data manifolds, and opens new directions for manifold-aware sampling, exploration, and generative modeling.

Emotion annotation is inherently subjective and cognitively demanding, producing signals that reflect diverse perceptions across annotators rather than a single ground truth. In continuous affect prediction, this variability is typically collapsed into point estimates such as the mean or median, discarding valuable information about annotator disagreement and uncertainty. In this work, we propose a distribution-aware framework that models annotation consensus using the Beta distribution. Instead of predicting a single affect value, models estimate the mean and standard deviation of the annotation distribution, which are transformed into valid Beta parameters through moment matching. This formulation enables the recovery of higher-order distributional descriptors, including skewness, kurtosis, and quantiles, in closed form. As a result, the model captures not only the central tendency of emotional perception but also variability, asymmetry, and uncertainty in annotator responses. We evaluate the proposed approach on the SEWA and RECOLA datasets using multimodal features. Experimental results show that Beta-based modelling produces predictive distributions that closely match the empirical annotator distributions while achieving competitive performance with conventional regression approaches. These findings highlight the importance of modelling annotation uncertainty in affective computing and demonstrate the potential of distribution-aware learning for subjective signal analysis.

Predicting affect in unconstrained environments remains a fundamental challenge in human-centered AI. While deep neural embeddings dominate contemporary approaches, they often lack interpretability and limit expert-driven refinement. We propose a novel framework that uses Language Models (LMs) as semantic context conditioners over handcrafted affect descriptors to model changes in Valence and Arousal. Our approach begins with interpretable facial geometry and acoustic features derived from structured domain knowledge. These features are transformed into symbolic natural-language descriptions encoding their affective implications. A pretrained LM processes these descriptions to generate semantic context embeddings that act as high-level priors over affective dynamics. Unlike end-to-end black-box pipelines, our framework preserves feature transparency while leveraging the contextual abstraction capabilities of LMs. We evaluate the proposed method on the Aff-Wild2 and SEWA datasets for affect change prediction. Experimental results show consistent improvements in accuracy for both Valence and Arousal compared to handcrafted-only and deep-embedding baselines. Our findings demonstrate that semantic conditioning enables interpretable affect modelling without sacrificing predictive performance, offering a transparent and computationally efficient alternative to fully end-to-end architectures

LLMs used for code generation are typically guided by engineering constraints--technology choices, dependency restrictions, and architectural patterns--expressed in verbose natural language. We investigate whether compact, structured constraint headers can reduce prompt token consumption without degrading constraint compliance. Across six experimental rounds spanning 11 models, 16 benchmark tasks, and over 830 LLM invocations, we find that compact headers reduce constraint-portion tokens by approximately 71% and full-prompt tokens by 25--30%, replicated across three independent rounds. However, we detect no statistically significant differences in constraint satisfaction rate (CSR) across three encoding forms or four propagation modes; observed effect sizes are negligible (Cliff's $δ$ < 0.01, 95% CI spanning $\pm$2.6 percentage points). This null pattern holds across two models from different capability tiers. A supplementary experiment with four non-CSS tasks provides additional cross-domain support for the encoding null result. The largest observed sources of compliance variance are constraint type ($Δ$ = 9 percentage points between normal and counter-intuitive constraints) and task domain: counter-intuitive constraints opposing model defaults fail at 10--100%, while conventional constraints achieve 99%+ compliance regardless of encoding. Model self-assessments systematically overestimate compliance relative to rule-based scoring, revealing a gap between constraint understanding and execution. Under the tested conditions, the primary benefit of compact constraint encoding is token reduction rather than compliance improvement, and engineering effort toward compliance is better directed at constraint design than prompt formatting.

Mixture proportion estimation (MPE) aims to estimate class priors from unlabeled data. This task is a critical component in weakly supervised learning, such as PU learning, learning with label noise, and domain adaptation. Existing MPE methods rely on the \textit{irreducibility} assumption or its variant for identifiability. In this paper, we propose novel assumptions based on conditional independence (CI) given the class label, which ensure identifiability even when irreducibility does not hold. We develop method of moments estimators under these assumptions and analyze their asymptotic properties. Furthermore, we present weakly-supervised kernel tests to validate the CI assumptions, which are of independent interest in applications such as causal discovery and fairness evaluation. Empirically, we demonstrate the improved performance of our estimators compared with existing methods and that our tests successfully control both type I and type II errors.\label{key}

We present a comprehensive adoption snapshot of the leading open language models and who is building them, focusing on the ~1.5K mainline open models from the likes of Alibaba's Qwen, DeepSeek, Meta's Llama, that are the foundation of an ecosystem crucial to researchers, entrepreneurs, and policy advisors. We document a clear trend where Chinese models overtook their counterparts built in the U.S. in the summer of 2025 and subsequently widened the gap over their western counterparts. We study a mix of Hugging Face downloads and model derivatives, inference market share, performance metrics and more to make a comprehensive picture of the ecosystem.

Corpus linguistics has traditionally relied on human researchers to formulate hypotheses, construct queries, and interpret results - a process demanding specialized technical skills and considerable time. We propose Agent-Driven Corpus Linguistics, an approach in which a large language model (LLM), connected to a corpus query engine via a structured tool-use interface, takes over the investigative cycle: generating hypotheses, querying the corpus, interpreting results, and refining analysis across multiple rounds. The human researcher sets direction and evaluates final output. Unlike unconstrained LLM generation, every finding is anchored in verifiable corpus evidence. We treat this not as a replacement for the corpus-based/corpus-driven distinction but as a complementary dimension: it concerns who conducts the inquiry, not the epistemological relationship between theory and data. We demonstrate the framework by linking an LLM agent to a CQP-indexed Gutenberg corpus (5 million tokens) via the Model Context Protocol (MCP). Given only "investigate English intensifiers," the agent identified a diachronic relay chain (so+ADJ > very > really), three pathways of semantic change (delexicalization, polarity fixation, metaphorical constraint), and register-sensitive distributions. A controlled baseline experiment shows that corpus grounding contributes quantification and falsifiability that the model cannot produce from training data alone. To test external validity, the agent replicated two published studies on the CLMET corpus (40 million tokens) - Claridge (2025) and De Smet (2013) - with close quantitative agreement. Agent-driven corpus research can thus produce empirically grounded findings at machine speed, lowering the technical barrier for a broader range of researchers.

As the worlds second most consumed beverage after water, tea is not just a cultural staple but a global economic force of profound scale and influence. More than a mere drink, it represents a quiet negotiation between nature, culture, and the human desire for a moment of reflection. So, the precise identification and detection of tea leaf disease is crucial. With this goal, we have evaluated several Convolutional Neural Networks (CNN) models, among them three shows noticeable performance including DenseNet201, MobileNetV2, InceptionV3 on the teaLeafBD dataset. teaLeafBD dataset contains seven classes, six disease classes and one healthy class, collected under various field conditions reflecting real world challenges. Among the CNN models, DenseNet201 has achieved the highest test accuracy of 99%. In order to enhance the model reliability and interpretability, we have implemented Gradient weighted Class Activation Mapping (Grad CAM), occlusion sensitivity analysis and adversarial training techniques to increase the noise resistance of the model. Finally, we have developed a prototype in order to leverage the models capabilities on real life agriculture. This paper illustrates the deep learning models capabilities to classify the disease in real life tea leaf disease detection and management.

We propose a geometric framework for longitudinal multi-parametric MRI analysis based on patient-specific energy modelling in sequence space. Rather than operating on images with spatial networks, each voxel is represented by its multi-sequence intensity vector ($T1$, $T1c$, $T2$, FLAIR, ADC), and a compact implicit neural representation is trained via denoising score matching to learn an energy function $E_θ(\mathbf{u})$ over $\mathbb{R}^d$ from a single baseline scan. The learned energy landscape provides a differential-geometric description of tissue regimes without segmentation labels. Local minima define tissue basins, gradient magnitude reflects proximity to regime boundaries, and Laplacian curvature characterises local constraint structure. Importantly, this baseline energy manifold is treated as a fixed geometric reference: it encodes the set of contrast combinations observed at diagnosis and is not retrained at follow-up. Longitudinal assessment is therefore formulated as evaluation of subsequent scans relative to this baseline geometry. Rather than comparing anatomical segmentations, we analyse how the distribution of MRI sequence vectors evolves under the baseline energy function. In a paediatric case with later recurrence, follow-up scans show progressive deviation in energy and directional displacement in sequence space toward the baseline tumour-associated regime before clear radiological reappearance. In a case with stable disease, voxel distributions remain confined to established low-energy basins without systematic drift. The presented cases serve as proof-of-concept that patient-specific energy manifolds can function as geometric reference systems for longitudinal mpMRI analysis without explicit segmentation or supervised classification, providing a foundation for further investigation of manifold-based tissue-at-risk tracking in neuro-oncology.

We investigate the feasibility of real-time 3D Gaussian Splatting (3DGS) rasterisation on edge clients with varying Gaussian splat counts and GPU computational budgets. Instead of evaluating multiple physical devices, we adopt an emulation-based approach that approximates different GPU capability tiers on a single high-end GPU. By systematically under-clocking the GPU core frequency and applying power caps, we emulate a controlled range of floating-point performance levels that approximate different GPU capability tiers. At each point in this range, we measure frame rate, runtime behaviour, and power consumption across scenes of varying complexity, pipelines, and optimisations, enabling analysis of power-performance relationships such as FPS-power curves, energy per frame, and performance per watt. This method allows us to approximate the performance envelope of a diverse class of GPUs, from embedded and mobile-class devices to high-end consumer-grade systems. Our objective is to explore the practical lower bounds of client-side 3DGS rasterisation and assess its potential for deployment in energy-constrained environments, including standalone headsets and thin clients. Through this analysis, we provide early insights into the performance-energy trade-offs that govern the viability of edge-deployed 3DGS systems.

Calibration is central to reliable semantic uncertainty quantification, yet prior work has largely focused on discrimination, neglecting calibration. As calibration and discrimination capture distinct aspects of uncertainty, focusing on discrimination alone yields an incomplete picture. We address this gap by systematically evaluating both aspects across a broad set of confidence measures. We show that current approaches, particularly fixed-temperature heuristics, produce systematically miscalibrated and poorly discriminative semantic confidence distributions. We demonstrate that optimising a single scalar temperature, which, we argue, provides a suitable inductive bias, is a surprisingly simple yet effective solution. Our exhaustive evaluation confirms that temperature scaling consistently improves semantic calibration, discrimination, and downstream entropy, outperforming both heuristic baselines and more expressive token-level recalibration methods on question-answering tasks.

Decision-making in military aviation Prognostics and Health Management (PHM) faces significant challenges due to the "curse of dimensionality" in large-scale fleet operations, combined with sparse feedback and stochastic mission profiles. To address these issues, this paper proposes Smart Commander, a novel Hierarchical Reinforcement Learning (HRL) framework designed to optimize sequential maintenance and logistics decisions. The framework decomposes the complex control problem into a two-tier hierarchy: a strategic General Commander manages fleet-level availability and cost objectives, while tactical Operation Commanders execute specific actions for sortie generation, maintenance scheduling, and resource allocation. The proposed approach is validated within a custom-built, high-fidelity discrete-event simulation environment that captures the dynamics of aircraft configuration and support logistics.By integrating layered reward shaping with planning-enhanced neural networks, the method effectively addresses the difficulty of sparse and delayed rewards. Empirical evaluations demonstrate that Smart Commander significantly outperforms conventional monolithic Deep Reinforcement Learning (DRL) and rule-based baselines. Notably, it achieves a substantial reduction in training time while demonstrating superior scalability and robustness in failure-prone environments. These results highlight the potential of HRL as a reliable paradigm for next-generation intelligent fleet management.

Bayesian filtering and smoothing for high-dimensional nonlinear dynamical systems are fundamental yet challenging problems in many areas of science and engineering. In this work, we propose AFSF, a unified amortized framework for filtering and smoothing with conditional normalizing flows. The core idea is to encode each observation history into a fixed-dimensional summary statistic and use this shared representation to learn both a forward flow for the filtering distribution and a backward flow for the backward transition kernel. Specifically, a recurrent encoder maps each observation history to a fixed-dimensional summary statistic whose dimension does not depend on the length of the time series. Conditioned on this shared summary statistic, the forward flow approximates the filtering distribution, while the backward flow approximates the backward transition kernel. The smoothing distribution over an entire trajectory is then recovered by combining the terminal filtering distribution with the learned backward flow through the standard backward recursion. By learning the underlying temporal evolution structure, AFSF also supports extrapolation beyond the training horizon. Moreover, by coupling the two flows through shared summary statistics, AFSF induces an implicit regularization across latent state trajectories and improves trajectory-level smoothing. In addition, we develop a flow-based particle filtering variant that provides an alternative filtering procedure and enables ESS-based diagnostics when explicit model factors are available. Numerical experiments demonstrate that AFSF provides accurate approximations of both filtering distributions and smoothing paths.

Although visual foundation models like DINOv2 provide state-of-the-art performance as feature extractors, their complex, high-dimensional representations create substantial hurdles for interpretability. This work proposes DINO-QPM, which converts these powerful but entangled features into contrastive, class-independent representations that are interpretable by humans. DINO-QPM is a lightweight interpretability adapter that pursues globally interpretable image classification, adapting the Quadratic Programming Enhanced Model (QPM) to operate on strictly frozen DINO backbones. While classification with visual foundation models typically relies on the \texttt{CLS} token, we deliberately diverge from this standard. By leveraging average-pooling, we directly connect the patch embeddings to the model's features and therefore enable spatial localisation of DINO-QPM's globally interpretable features within the input space. Furthermore, we apply a sparsity loss to minimise spatial scatter and background noise, ensuring that explanations are grounded in relevant object parts. With DINO-QPM we make the level of interpretability of QPM available as an adapter while exceeding the accuracy of DINOv2 linear probe. Evaluated through an introduced Plausibility metric and other interpretability metrics, extensive experiments demonstrate that DINO-QPM is superior to other applicable methods for frozen visual foundation models in both classification accuracy and explanation quality.

Reinforcement learning for Large Language Model agents is often hindered by sparse rewards in multi-step reasoning tasks. Existing approaches like Group Relative Policy Optimization treat sampled trajectories as independent chains, assigning uniform credit to all steps in each chain and ignoring the existence of critical steps that may disproportionally impact reasoning outcome. In this paper, we propose T-STAR(Tree-structured Self-Taught Agent Rectification), a framework that recovers the latent correlated reward structure across seemingly independent trajectories. Specifically, we consolidate trajectories into a unified Cognitive Tree by identifying and merging functionally similar steps/nodes. It enables an Introspective Valuation mechanism that back-propagates trajectory-level rewards through the tree to obtain a new notion of variance-reduced relative advantage at step-level. Using the Cognitive Tree, we also develop In-Context Thought Grafting to synthesize corrective reasoning by contrasting successful and failed branches at critical divergence points/steps. Our proposed Surgical Policy Optimization then capitalizes on the rich policy gradient information concentrated at these critical points/steps through a Bradley-Terry type of surgical loss. Extensive experiments across embodied, interactive, reasoning, and planning benchmarks demonstrate that T-STAR achieves consistent improvements over strong baselines, with gains most pronounced on tasks requiring extended reasoning chains.

We study the problem of generating synthetic time series that reproduce both marginal distributions and temporal dynamics, a central challenge in financial machine learning. Existing approaches typically fail to jointly model drift and stochastic volatility, as diffusion-based methods fix the volatility while martingale transport models ignore drift. We introduce the Schrödinger-Bass Bridge for Time Series (SBBTS), a unified framework that extends the Schrödinger-Bass formulation to multi-step time series. The method constructs a diffusion process that jointly calibrates drift and volatility and admits a tractable decomposition into conditional transport problems, enabling efficient learning. Numerical experiments on the Heston model demonstrate that SBBTS accurately recovers stochastic volatility and correlation parameters that prior SchrödingerBridge methods fail to capture. Applied to S&P 500 data, SBBTS-generated synthetic time series consistently improve downstream forecasting performance when used for data augmentation, yielding higher classification accuracy and Sharpe ratio compared to real-data-only training. These results show that SBBTS provides a practical and effective framework for realistic time series generation and data augmentation in financial applications.

Multimodal imaging analysis often relies on joint latent representations, yet these approaches rarely define what information is shared versus modality-specific. Clarifying this distinction is clinically relevant, as it delineates the irreducible contribution of each modality and informs rational acquisition strategies. We propose a subspace decomposition framework that reframes multimodal fusion as a problem of orthogonal subspace separation rather than translation. We decompose Prostate-Specific Membrane Antigen (PSMA) PET uptake into an MRI-explainable physiological envelope and an orthogonal residual reflecting signal components not expressible within the MRI feature manifold. Using multiparametric MRI, we train an intensity-based, non-spatial implicit neural representation (INR) to map MRI feature vectors to PET uptake. We introduce a projection-based regularization using singular value decomposition to penalize residual components lying within the span of the MRI feature manifold. This enforces mathematical orthogonality between tissue-level physiological properties (structure, diffusion, perfusion) and intracellular PSMA expression. Tested on 13 prostate cancer patients, the model demonstrates that residual components spanned by MRI features are absorbed into the learned envelope, while the orthogonal residual is largest in tumour regions. This indicates that PSMA PET contains signal components not recoverable from MRI-derived physiological descriptors. The resulting decomposition provides a structured characterization of modality complementarity grounded in representation geometry rather than image translation.

Emerging computation-intensive applications impose stringent latency requirements on resource-constrained mobile devices. Mobile Edge Computing (MEC) addresses this challenge through task offloading. However, designing effective policies remains difficult due to dynamic task arrivals, time-varying channels, and the spatio-temporal coupling of server queues. Conventional heuristics lack adaptability, while Deep Reinforcement Learning (DRL) suffers from limited generalization and architectural rigidity, requiring retraining when network topology changes. Although Large Language Models (LLMs) offer semantic reasoning capabilities, standard Supervised Fine-Tuning (SFT) yields myopic policies that greedily minimize immediate latency without accounting for long-term system evolution. To address these limitations, we propose COMLLM, a generative framework that enables foresighted decision-making in MEC systems. COMLLM integrates Group Relative Policy Optimization (GRPO) with a Look-Ahead Collaborative Simulation (LACS) mechanism, which performs multi-step Monte Carlo rollouts while jointly modeling server queue dynamics. By incorporating these rollouts into the reward design, the framework captures the long-term impact of current decisions on future system states. Experimental results demonstrate that COMLLM achieves near-optimal latency and improved load-balancing fairness. Notably, it exhibits zero-shot topological scalability, allowing a model trained on small-scale networks to generalize to larger, unseen topologies without retraining, outperforming SFT, DRL, and heuristic baselines.

Large language models produce repetitive output when prompted independently across many batches, a phenomenon we term cross-batch mode collapse: the progressive loss of output diversity when a language model is prompted repeatedly without access to its prior generations. Practitioners have long mitigated this with ad hoc deduplication and seed rotation, but no principled framework exists. We introduce Dynamic Context Evolution (DCE), comprising three mechanisms: (1) verbalized tail sampling (the model labels each idea with a guess about how obvious it is, and obvious ideas are discarded), which filters high-probability candidates via model self-assessment; (2) semantic memory, which maintains a persistent embedding index to reject near-duplicates across batches; and (3) adaptive prompt evolution, which reconstructs the generation prompt each batch using memory state and rotating diversity strategies. In experiments across three domains (sustainable packaging concepts, educational exam questions, and creative writing prompts) and two model families (gpt-5-mini and claude-haiku-4-5), a component ablation across 2-3 random seeds per method shows that DCE achieves 0.0 +/- 0.0% collapse versus 5.6 +/- 2.0% for naive prompting, while producing 17-18 HDBSCAN clusters per seed versus naive's volatile 2-17, indicating reliably richer conceptual structure. These results are validated with an independent embedding model (all-MiniLM-L6-v2) and hold across sensitivity sweeps of the VTS threshold tau and dedup threshold delta. Deduplication and prompt evolution are individually insufficient but jointly effective, at approximately $0.50 per 1,000 candidates using only standard API calls, with no fine-tuning or custom architectures required.

We introduce Lumbermark, a robust divisive clustering algorithm capable of detecting clusters of varying sizes, densities, and shapes. Lumbermark iteratively chops off large limbs connected by protruding segments of a dataset's mutual reachability minimum spanning tree. The use of mutual reachability distances smoothens the data distribution and decreases the influence of low-density objects, such as noise points between clusters or outliers at their peripheries. The algorithm can be viewed as an alternative to HDBSCAN that produces partitions with user-specified sizes. A fast, easy-to-use implementation of the new method is available in the open-source 'lumbermark' package for Python and R. We show that Lumbermark performs well on benchmark data and hope it will prove useful to data scientists and practitioners across different fields.

This paper focuses on energy savings in downlink operation of cell-free massive MIMO (CF mMIMO) networks under dynamic traffic conditions. We propose a multi-agent deep reinforcement learning (MADRL) algorithm that enables each access point (AP) to autonomously control antenna re-configuration and advanced sleep mode (ASM) selection. After the training process, the proposed framework operates in a fully distributed manner, eliminating the need for centralized control and allowing each AP to dynamically adjust to real-time traffic fluctuations. Simulation results show that the proposed algorithm reduces power consumption (PC) by 56.23% compared to systems without any energy-saving scheme and by 30.12% relative to a non-learning mechanism that only utilizes the lightest sleep mode, with only a slight increase in drop ratio. Moreover, compared to the widely used deep Q-network (DQN) algorithm, it achieves a similar PC level but with a significantly lower drop ratio.

Child Sexual Abuse Imagery (CSAI) classification is an important yet challenging problem for computer vision research due to the strict legal and ethical restrictions that prevent the public sharing of CSAI datasets. This limitation hinders reproducibility and slows progress in developing automated methods. In this work, we introduce CSA-Graphs, a privacy-preserving structural dataset. Instead of releasing the original images, we provide structural representations that remove explicit visual content while preserving contextual information. CSA-Graphs includes two complementary graph-based modalities: scene graphs describing object relationships and skeleton graphs encoding human pose. Experiments show that both representations retain useful information for classifying CSAI, and that combining them further improves performance. This dataset enables broader research on computer vision methods for child safety while respecting legal and ethical constraints.

Numerical simulation of wave propagation and run-up is a cornerstone of coastal engineering and tsunami hazard assessment. However, applying these forward models to inverse problems, such as bathymetry estimation, source inversion, and structural optimization, remains notoriously difficult due to the rigidity and high computational cost of deriving discrete adjoints. In this paper, we introduce AegirJAX, a fully differentiable hydrodynamic solver based on the depth-integrated, non-hydrostatic shallow-water equations. By implementing the solver entirely within a reverse-mode automatic differentiation framework, AegirJAX treats the time-marching physics loop as a continuous computational graph. We demonstrate the framework's versatility across a suite of scientific machine learning tasks: (1) discovering regime-specific neural corrections for model misspecifications in highly dispersive wave propagation; (2) performing continuous topology optimization for breakwater design; (3) training recurrent neural networks in-the-loop for active wave cancellation; and (4) inverting hidden bathymetry and submarine landslide kinematics directly from downstream sensor data. The proposed differentiable paradigm fundamentally blurs the line between forward simulation and inverse optimization, offering a unified, end-to-end framework for coastal hydrodynamics.

Predicting vehicle trajectories plays an important role in autonomous driving and ITS applications. Although multiple deep learning algorithms are devised to predict vehicle trajectories, their reliant on specific graph structure (e.g., Graph Neural Network) or explicit intention labeling limit their flexibilities. In this study, we propose a pure Transformer-based network with multiple modals considering their neighboring vehicles. Two separate tracks are employed. One track focuses on predicting the trajectories while the other focuses on predicting the likelihood of each intention considering neighboring vehicles. Study finds that the two track design can increase the performance by separating spatial module from the trajectory generating module. Also, we find the the model can learn an ordered group of trajectories by predicting residual offsets among K trajectories.

This article presents DDP-SA, a scalable privacy-preserving federated learning framework that jointly leverages client-side local differential privacy (LDP) and full-threshold additive secret sharing (ASS) for secure aggregation. Unlike existing methods that rely solely on differential privacy or on secure multi-party computation (MPC), DDP-SA integrates both techniques to deliver stronger end-to-end privacy guarantees while remaining computationally practical. The framework introduces a two-stage protection mechanism: clients first perturb their local gradients with calibrated Laplace noise, then decompose the noisy gradients into additive secret shares that are distributed across multiple intermediate servers. This design ensures that (i) no single compromised server or communication channel can reveal any information about individual client updates, and (ii) the parameter server reconstructs only the aggregated noisy gradient, never any client-specific contribution. Extensive experiments show that DDP-SA achieves substantially higher model accuracy than standalone LDP while providing stronger privacy protection than MPC-only approaches. The proposed framework scales linearly with the number of participants and offers a practical, privacy-preserving solution for federated learning applications with controllable computational and communication overhead.

Large Language Models (LLMs) have been shown to contain biases in the process of integrating conflicting information when answering questions. Here we ask whether such biases also exist with respect to which language is used for each conflicting piece of information. To answer this question, we extend the conflicting needles in a haystack paradigm to a multilingual setting and perform a comprehensive set of evaluations with naturalistic news domain data in five different languages, for a range of multilingual LLMs of different sizes. We find that all LLMs tested, including GPT-5.2, ignore the conflict and confidently assert only one of the possible answers in the large majority of cases. Furthermore, there is a consistent bias across models in which languages are preferred, with a general bias against Russian and, for the longest context lengths, in favor of Chinese. Both of these patterns are consistent between models trained inside and outside of mainland China, though somewhat stronger in the former category.

In semantic segmentation, the creation of pixel-level labels for training data incurs significant costs. To address this problem, semi-supervised learning, which utilizes a small number of labeled images alongside unlabeled images to enhance the performance, has gained attention. A conventional semi-supervised learning method, ClassMix, pastes class labels predicted from unlabeled images onto other images. However, since ClassMix performs operations using pseudo-labels obtained from unlabeled images, there is a risk of handling inaccurate labels. Additionally, there is a gap in data quality between labeled and unlabeled images, which can impact the feature maps. This study addresses these two issues. First, we propose a method where class labels from labeled images, along with the corresponding image regions, are pasted onto unlabeled images and their pseudo-labeled images. Second, we introduce a method that trains the model to make predictions on unlabeled images more similar to those on labeled images. Experiments on the Chase and COVID-19 datasets demonstrated an average improvement of 2.07% in mIoU compared to conventional semi-supervised learning methods.

In the human-AI collaboration area, the context formed naturally through multi-turn interactions is typically flattened into a chronological sequence and treated as a fixed whole in subsequent reasoning, with no mechanism for dynamic organization and management along the collaboration workflow. Yet these contexts differ substantially in lifecycle, structural hierarchy, and relevance. For instance, temporary or abandoned exchanges and parallel topic threads persist in the limited context window, causing interference and even conflict. Meanwhile, users are largely limited to influencing context indirectly through input modifications (e.g., corrections, references, or ignoring), leaving their control neither explicit nor verifiable. To address this, we propose Mixed-Initiative Context, which reconceptualizes the context formed across multi-turn interactions as an explicit, structured, and manipulable interactive object. Under this concept, the structure, scope, and content of context can be dynamically organized and adjusted according to task needs, enabling both humans and AI to actively participate in context construction and regulation. To explore this concept, we implement Contextify as a probe system and conduct a user study examining users' context management behaviors, attitudes toward AI initiative, and overall collaboration experience. We conclude by discussing the implications of this concept for the HCI community.

Current operational Earth Observation (EO) services, including the Copernicus Emergency Management Service (CEMS), the European Forest Fire Information System (EFFIS), and the Copernicus Land Monitoring Service (CLMS), rely primarily on ground-based processing pipelines. While these systems provide mature large-scale information products, they remain constrained by downlink latency, bandwidth limitations, and limited capability for autonomous observation prioritisation. The International Report for an Innovative Defence of Earth (IRIDE) programme is a national Earth observation initiative led by the Italian government to support public authorities through timely, objective information derived from spaceborne data. Rather than a single constellation, IRIDE is designed as a constellation of constellations, integrating heterogeneous sensing technologies within a unified service-oriented architecture. Within this framework, Hawk for Earth Observation (HEO) enables onboard generation of data products, allowing information extraction earlier in the processing chain. This paper examines the limitations of ground-only architectures and evaluates the added value of onboard processing at the operational service level. The IRIDE burnt-area mapping service is used as a representative case study to demonstrate how onboard intelligence can support higher spatial detail (sub-three-metre ground sampling distance), smaller detectable events (minimum mapping unit of three hectares), and improved system responsiveness. Rather than replacing existing Copernicus services, the IRIDE HEO capability is positioned as a complementary layer providing image-driven pre-classification to support downstream emergency and land-management workflows. This work highlights the operational value of onboard intelligence for emerging low-latency EO service architectures.

Peer review plays a central role in the NLP publication process, but is susceptible to various biases. Here, we study language-of-study (LoS) bias: the tendency for reviewers to evaluate a paper differently based on the language(s) it studies, rather than its scientific merit. Despite being explicitly flagged in reviewing guidelines, such biases are poorly understood. Prior work treats such comments as part of broader categories of weak or unconstructive reviews without defining them as a distinct form of bias. We present the first systematic characterization of LoS bias, distinguishing negative and positive forms, and introduce the human-annotated dataset LOBSTER (Language-Of-study Bias in ScienTific pEer Review) and a method achieving 87.37 macro F1 for detection. We analyze 15,645 reviews to estimate how negative and positive biases differ with respect to the LoS, and find that non-English papers face substantially higher bias rates than English-only ones, with negative bias consistently outweighing positive bias. Finally, we identify four subcategories of negative bias, and find that demanding unjustified cross-lingual generalization is the most dominant form. We publicly release all resources to support work on fairer reviewing practices in NLP and beyond.

We describe the Yale-DM-Lab system for the ArchEHR-QA 2026 shared task. The task studies patient-authored questions about hospitalization records and contains four subtasks (ST): clinician-interpreted question reformulation, evidence sentence identification, answer generation, and evidence-answer alignment. ST1 uses a dual-model pipeline with Claude Sonnet 4 and GPT-4o to reformulate patient questions into clinician-interpreted questions. ST2-ST4 rely on Azure-hosted model ensembles (o3, GPT-5.2, GPT-5.1, and DeepSeek-R1) combined with few-shot prompting and voting strategies. Our experiments show three main findings. First, model diversity and ensemble voting consistently improve performance compared to single-model baselines. Second, the full clinician answer paragraph is provided as additional prompt context for evidence alignment. Third, results on the development set show that alignment accuracy is mainly limited by reasoning. The best scores on the development set reach 88.81 micro F1 on ST4, 65.72 macro F1 on ST2, 34.01 on ST3, and 33.05 on ST1.

Catastrophic forgetting is not an engineering failure. It is a mathematical consequence of storing knowledge as global parameter superposition. Existing methods, such as regularization, replay, and frozen subnetworks, add external mechanisms to a shared-parameter substrate. None derives retention from the learning dynamics themselves. This paper introduces the Informational Buildup Framework (IBF), an alternative substrate for continual learning, based on the premise that information is the achievement of structural alignment rather than stored content. In IBF, two equations govern the dynamics: a Law of Motion that drives configuration toward higher coherence, and Modification Dynamics that persistently deform the coherence landscape in response to localized discrepancies. Memory, agency, and self-correction arise from these dynamics rather than being added as separate modules. We first demonstrate the full lifecycle in a transparent two-dimensional toy model, then validate across three domains: a controlled non-stationary world, chess evaluated independently by Stockfish, and Split-CIFAR-100 with a frozen ViT encoder. Across all three, IBF achieves replay-superior retention without storing raw data. We observe near-zero forgetting on CIFAR-100 (BT = -0.004), positive backward transfer in chess (+38.5 cp), and 43% less forgetting than replay in the controlled domain. In chess, the framework achieves a mean behavioral advantage of +88.9 +/- 2.8 cp under independent evaluation, exceeding MLP and replay baselines.

Activation-based steering can personalize large language models at inference time, but its effects in educational settings remain unclear. We study persona vectors for seven character traits in short-answer generation and automated scoring on the ASAP-SAS benchmark across three models spanning two architectures. Persona steering lowers answer quality overall, with much larger effects on open-ended English Language Arts (ELA) prompts than on factual science prompts; interpretive and argumentative tasks are up to 11x more sensitive. On the scoring side, we observe predictable valence-aligned calibration shifts: evil and impolite scorers grade more harshly, while good and optimistic scorers grade more leniently. ELA tasks are 2.5-3x more susceptible to scorer personalization than science tasks, and the Mixture-of-Experts model shows roughly 6x larger calibration shifts than the dense models. To our knowledge, this is the first study to systematically examine the effects of activation-steered persona traits in educational generation and scoring, and the results highlight the need for task-aware and architecture-aware calibration when deploying steered models in educational settings.

We present the Surveillance Forgery Image Test Range (SurFITR), a dataset for surveillance-style image forgery detection and localisation, in response to recent advances in open-access image generation models that raise concerns about falsifying visual evidence. Existing forgery models, trained on datasets with full-image synthesis or large manipulated regions in object-centric images, struggle to generalise to surveillance scenarios. This is because tampering in surveillance imagery is typically localised and subtle, occurring in scenes with varied viewpoints, small or occluded subjects, and lower visual quality. To address this gap, SurFITR provides a large collection of forensically valuable imagery generated via a multimodal LLM-powered pipeline, enabling semantically aware, fine-grained editing across diverse surveillance scenes. It contains over 137k tampered images with varying resolutions and edit types, generated using multiple image editing models. Extensive experiments show that existing detectors degrade significantly on SurFITR, while training on SurFITR yields substantial improvements in both in-domain and cross-domain performance. SurFITR is publicly available on GitHub.

Empathetic dialogue requires not only recognizing a user's emotional state but also making strategy-aware, context-sensitive decisions throughout response generation. However, the lack of a comprehensive empathy strategy framework, explicit task-aligned multi-stage reasoning, and high-quality strategy-aware data fundamentally limits existing approaches, preventing them from effectively modeling empathetic dialogue as a complex, multi-stage cognitive and decision-making process. To address these challenges, we propose STRIDE-ED, a STRategy-grounded, Interpretable, and DEep reasoning framework that models Empathetic Dialogue through structured, strategy-conditioned reasoning. To support effective learning, we develop a strategy-aware data refinement pipeline integrating LLM-based annotation, multi-model consistency-weighted evaluation, and dynamic sampling to construct high-quality training data aligned with empathetic strategies. Furthermore, we adopt a two-stage training paradigm that combines supervised fine-tuning with multi-objective reinforcement learning to better align model behaviors with target emotions, empathetic strategies, and response formats. Extensive experiments demonstrate that STRIDE-ED generalizes across diverse open-source LLMs and consistently outperforms existing methods on both automatic metrics and human evaluations.

Large language models often fail on tasks they seem to already understand. In our experiments, this appears to be less about missing knowledge and more about certain internal circuits not being strongly activated during inference. We explore Selective Neuron Amplification, which increases the influence of task relevant neurons without changing the model's parameters. The method works at inference time and does not permanently alter the model. SNA helps mainly when the model is uncertain, while having low effect when the model is already confident. This suggests that some model failures are due to weak activation rather than lack of capability.

Multi-objective bandits have attracted increasing attention because of their broad applicability and mathematical elegance, where the reward of each arm is a multi-dimensional vector rather than a scalar. This naturally introduces Pareto order relations and Pareto regret. A long-standing question in this area is whether performance is fundamentally harder to optimize because of this added complexity. A recent surprising result shows that, in the adversarial setting, Pareto regret is no larger than classical regret; however, in the stochastic setting, where the regret notion is different, the picture remains unclear. In fact, existing work suggests that Pareto regret in the stochastic case increases with the dimensionality. This controversial yet subtle phenomenon motivates our central question: \emph{are multi-objective bandits actually harder than single-objective ones?} We answer this question in full by showing that, in the stochastic setting, Pareto regret is in fact governed by the maximum sub-optimality gap \(g^\dagger\), and hence by the minimum marginal regret of order \(Ω(\frac{K\log T}{g^\dagger})\). We further develop a new algorithm that achieves Pareto regret of order \(O(\frac{K\log T}{g^\dagger})\), and is therefore optimal. The algorithm leverages a nested two-layer uncertainty quantification over both arms and objectives through upper and lower confidence bound estimators. It combines a top-two racing strategy for arm selection with an uncertainty-greedy rule for dimension selection. Together, these components balance exploration and exploitation across the two layers. We also conduct comprehensive numerical experiments to validate the proposed algorithm, showing the desired regret guarantee and significant gains over benchmark methods.

Do multilingual embedding models encode a language-general representation of proficiency? We investigate this by training linear and non-linear probes on hidden-state activations from Qwen3-Embedding (0.6B, 4B, 8B) to predict CEFR proficiency levels from learner texts across nine corpora and seven languages. We compare five probing architectures against a baseline trained on surface-level text features. Under in-distribution evaluation, probes achieve strong performance ($QWK\approx0.7$), substantially outperforming the surface baseline, with middle layers consistently yielding the best predictions. However, in cross-corpus evaluation performance collapses across all probe types and model sizes. Residual analysis reveals that out-of-distribution probes converge towards predicting uniformly distributed labels, indicating that the learned mappings capture corpus-specific distributional properties (topic, language, task type, rating methodology) rather than an abstract, transferable proficiency dimension. These results suggest that current multilingual embeddings do not straightforwardly encode language-general proficiency, with implications for representation-based approaches to proficiency-adaptive language technology.

In electronic health records (EHRs), clustering patients and distinguishing disease subtypes are key tasks to elucidate pathophysiology and aid clinical decision-making. However, clustering in healthcare informatics is still based on traditional methods, especially K-means, and has achieved limited success when applied to embedding representations learned by autoencoders as hybrid methods. This paper investigates the effectiveness of traditional, hybrid, and deep learning methods in heart failure patient cohorts using real EHR data from the All of Us Research Program. Traditional clustering methods perform robustly because deep learning approaches are specifically designed for image clustering, a task that differs substantially from the tabular EHR data setting. To address the shortcomings of deep clustering, we introduce an ensemble-based deep clustering approach that aggregates cluster assignments obtained from multiple embedding dimensions, rather than relying on a single fixed embedding space. When combined with traditional clustering in a novel ensemble framework, the proposed ensemble embedding for deep clustering delivers the best overall performance ranking across 14 diverse clustering methods and multiple patient cohorts. This paper underscores the importance of biological sex-specific clustering of EHR data and the advantages of combining traditional and deep clustering approaches over a single method.

Open-loop end-to-end neural motion planners have recently been proposed to improve motion planning for robotic manipulators. These methods enable planning directly from sensor observations without relying on a privileged collision checker during planning. However, many existing methods generate only a single path for a given workspace across different runs, and do not leverage their open-loop structure for inference-time optimization. To address this limitation, we introduce Flow Motion Policy, an open-loop, end-to-end neural motion planner for robotic manipulators that leverages the stochastic generative formulation of flow matching methods to capture the inherent multi-modality of planning datasets. By modeling a distribution over feasible paths, Flow Motion Policy enables efficient inference-time best-of-$N$ sampling. The method generates multiple end-to-end candidate paths, evaluates their collision status after planning, and executes the first collision-free solution. We benchmark the Flow Motion Policy against representative sampling-based and neural motion planning methods. Evaluation results demonstrate that Flow Motion Policy improves planning success and efficiency, highlighting the effectiveness of stochastic generative policies for end-to-end motion planning and inference-time optimization. Experimental evaluation videos are available via this \href{https://zh.engr.tamu.edu/wp-content/uploads/sites/310/2026/03/FMP-Website.mp4}{link}.

Assessing the effectiveness of REST API tests in black-box settings can be challenging due to the lack of access to source code coverage metrics and polyglot tech stack. We propose three metrics for capturing average, minimum, and maximum log coverage to handle the diverse test generation results and runtime behaviors over multiple runs. Using log coverage, we empirically evaluate three REST API test generation strategies, Evolutionary computing (EvoMaster v5.0.2), LLMs (Claude Opus 4.6 and GPT-5.2-Codex), and human-written Locust load tests, on Light-OAuth2 authorization microservice system. On average, Claude Opus 4.6 tests uncover 28.4% more unique log templates than human-written tests, whereas EvoMaster and GPT-5.2-Codex find 26.1% and 38.6% fewer, respectively. Next, we analyze combined log coverage to assess complementarity between strategies. Combining human-written tests with Claude Opus 4.6 tests increases total observed log coverage by 78.4% and 38.9% in human-written and Claude tests respectively. When combining Locust tests with EvoMaster the same increases are 30.7% and 76.9% and when using GPT-5.2-Codex 26.1% and 105.6%. This means that the generation strategies exercise largely distinct runtime behaviors. Our future work includes extending our study to multiple systems.

Offline reinforcement learning learns policies from fixed datasets without further environment interaction. A key challenge in this setting is epistemic uncertainty, arising from limited or biased data coverage, particularly when the behavior policy systematically avoids certain actions. This can lead to inaccurate value estimates and unreliable generalization. Ensemble-based methods like SAC-N mitigate this by conservatively estimating Q-values using the ensemble minimum, but they require large ensembles and often conflate epistemic with aleatoric uncertainty. To address these limitations, we propose a unified and generalizable framework that replaces discrete ensembles with compact uncertainty sets over Q-values. %We further introduce an Epinet based model that directly shapes the uncertainty sets to optimize the cumulative reward under the robust Bellman objective without relying on ensembles. We also introduce a benchmark for evaluating offline RL algorithms under risk-sensitive behavior policies, and demonstrate that our method achieves improved robustness and generalization over ensemble-based baselines across both tabular and continuous state domains.

While Large Language Models (LLMs) demonstrate remarkable reasoning capabilities, their potential for purpose-driven exploration in dynamic geo-spatial environments remains under-investigated. Existing Geo-Spatial Question Answering (GSQA) benchmarks predominantly focus on static retrieval, failing to capture the complexity of real-world planning that involves dynamic user locations and compound constraints. To bridge this gap, we introduce EVGeoQA, a novel benchmark built upon Electric Vehicle (EV) charging scenarios that features a distinct location-anchored and dual-objective design. Specifically, each query in EVGeoQA is explicitly bound to a user's real-time coordinate and integrates the dual objectives of a charging necessity and a co-located activity preference. To systematically assess models in such complex settings, we further propose GeoRover, a general evaluation framework based on a tool-augmented agent architecture to evaluate the LLMs' capacity for dynamic, multi-objective exploration. Our experiments reveal that while LLMs successfully utilize tools to address sub-tasks, they struggle with long-range spatial exploration. Notably, we observe an emergent capability: LLMs can summarize historical exploration trajectories to enhance exploration efficiency. These findings establish EVGeoQA as a challenging testbed for future geo-spatial intelligence. The dataset and prompts are available at https://github.com/Hapluckyy/EVGeoQA/.

This paper presents an indirect data-driven output feedback controller synthesis for nonlinear systems, leveraging Structured State-space Models (SSMs) as surrogate models. SSMs have emerged as a compelling alternative in modelling time-series data and dynamical systems. They can capture long-term dependencies while maintaining linear computational complexity with respect to the sequence length, in comparison to the quadratic complexity of Transformer-based architectures. The contributions of this work are threefold. We provide the first analysis of controllability and observability of SSMs, which leads to scalable control design via Linear Matrix Inequalities (LMIs) that leverage contraction theory. Moreover, a separation principle for SSMs is established, enabling the independent design of observers and state-feedback controllers while preserving the exponential stability of the closed-loop system. The effectiveness of the proposed framework is demonstrated through a numerical example, showcasing nonlinear system identification and the synthesis of an output feedback controller.

Bilingual speakers show cross-lingual activation during reading, especially for words with shared surface form. Cognates (friends) typically lead to facilitation, whereas interlingual homographs (false friends) cause interference or no effect. We examine whether cross-lingual activation in bilingual language models mirrors these patterns. We train Dutch-English causal Transformers under four vocabulary-sharing conditions that manipulate whether (false) friends receive shared or language-specific embeddings. Using psycholinguistic stimuli from bilingual reading studies, we evaluate the models through surprisal and embedding similarity analyses. The models largely maintain language separation, and cross-lingual effects arise primarily when embeddings are shared. In these cases, both friends and false friends show facilitation relative to controls. Regression analyses reveal that these effects are mainly driven by frequency rather than consistency in form-meaning mapping. Only when just friends share embeddings are the qualitative patterns of bilinguals reproduced. Overall, bilingual language models capture some cross-linguistic activation effects. However, their alignment with human processing seems to critically depend on how lexical overlap is encoded, possibly limiting their explanatory adequacy as models of bilingual reading.

We present the SemEval-2026 shared task on Dimensional Aspect-Based Sentiment Analysis (DimABSA), which improves traditional ABSA by modeling sentiment along valence-arousal (VA) dimensions rather than using categorical polarity labels. To extend ABSA beyond consumer reviews to public-issue discourse (e.g., political, energy, and climate issues), we introduce an additional task, Dimensional Stance Analysis (DimStance), which treats stance targets as aspects and reformulates stance detection as regression in the VA space. The task consists of two tracks: Track A (DimABSA) and Track B (DimStance). Track A includes three subtasks: (1) dimensional aspect sentiment regression, (2) dimensional aspect sentiment triplet extraction, and (3) dimensional aspect sentiment quadruplet extraction, while Track B includes only the regression subtask for stance targets. We also introduce a continuous F1 (cF1) metric to jointly evaluate structured extraction and VA regression. The task attracted more than 400 participants, resulting in 112 final submissions and 42 system description papers. We report baseline results, discuss top-performing systems, and analyze key design choices to provide insights into dimensional sentiment analysis at the aspect and stance-target levels. All resources are available on our GitHub repository.

Electronic Control Units (ECUs) have played a pivotal role in transforming motorcars of yore into the modern vehicles we see on our roads today. They actively regulate the actuation of individual components and thus determine the characteristics of the whole system. In this, the behavior of the control functions heavily depends on their calibration parameters which engineers traditionally design by hand. This is taking place in an environment of rising customer expectations and steadily shorter product development cycles. At the same time, legislative requirements are increasing while emission standards are getting stricter. Considering the number of vehicle variants on top of all that, the conventional method is losing its practical and financial viability. Prior work has already demonstrated that optimal control functions can be automatically developed with reinforcement learning (RL); since the resulting functions are represented by artificial neural networks, they lack explainability, a circumstance which renders them challenging to employ in production vehicles. In this article, we present an explainable approach to automating the calibration process using residual RL which follows established automotive development principles. Its applicability is demonstrated by means of a map-based air path controller in a series control unit using a hardware-in-the-loop (HiL) platform. Starting with a sub-optimal map, the proposed methodology quickly converges to a calibration which closely resembles the reference in the series ECU. The results prove that the approach is suitable for the industry where it leads to better calibrations in significantly less time and requires virtually no human intervention

Existing Indonesian sentiment analysis models classify text in isolation, ignoring the topical context that often determines whether a statement is positive, negative, or neutral. We introduce IndoBERT-Sentiment, a context-conditioned sentiment classifier that takes both a topical context and a text as input, producing sentiment predictions grounded in the topic being discussed. Built on IndoBERT Large (335M parameters) and trained on 31,360 context-text pairs labeled across 188 topics, the model achieves an F1 macro of 0.856 and accuracy of 88.1%. In a head-to-head evaluation against three widely used general-purpose Indonesian sentiment models on the same test set, IndoBERT-Sentiment outperforms the best baseline by 35.6 F1 points. We show that context-conditioning, previously demonstrated for relevancy classification, transfers effectively to sentiment analysis and enables the model to correctly classify texts that are systematically misclassified by context-free approaches.

We give a computer-assisted counterexample to the open question, posed by Rudin, Schapire, and Daubechies in COLT 2012, of whether exhaustive AdaBoost always converges to a finite cycle. The construction is based on a block-product gadget whose two factors share an exact period-2 orbit for their 5-step branch maps, but whose linearized return maps have dominant eigenvalues with an irrational logarithmic ratio. This irrationality forces the burst-winner sequence to have an irrational asymptotic frequency, precluding eventual periodicity. All assertions are certified by exact rational arithmetic. This work was developed in collaboration with GPT-5.4 Pro and Claude Opus 4.6.

Sales dialogues require multi-turn, goal-directed persuasion under asymmetric incentives, which makes them a challenging setting for large language models (LLMs). Yet existing dialogue benchmarks rarely measure deal progression and outcomes. We introduce SalesLLM, a bilingual (ZH/EN) benchmark derived from realistic applications covering Financial Services and Consumer Goods, built from 30,074 scripted configurations and 1,805 curated multi-turn scenarios with controllable difficulty and personas. We propose a fully automatic evaluation pipeline that combines (i) an LLM-based rater for sales-process progress, and (ii) fine-tuned BERT classifiers for end-of-dialogue buying intent. To improve simulation fidelity, we train a user model, CustomerLM, with SFT and DPO on 8,000 crowdworker-involved sales conversations, reducing role inversion from 17.44% (GPT-4o) to 8.8%. SalesLLM scores correlate strongly with expert human ratings (Pearson r=0.98). Experiments across 15 mainstream LLMs reveal substantial variability: top-performance LLMs are competitive with human-level performance while the less capable ones are worse than human. SalesLLM serves as a scalable benchmark for developing and evaluating outcome-oriented sales agents.

Most research in planning focuses on generating a plan to achieve a desired set of goals. However, a goal specification can also be used to encode a property that should never hold, allowing a planner to identify a trace that would reach a flawed state. In such cases, the objective may shift to modifying the planning task to ensure that the flawed state is never reached-in other words, to make the planning task unsolvable. In this paper we introduce planning task shielding: the problem of detecting and repairing flaws in planning tasks. We propose $allmin$, an optimal algorithm that solves these tasks by minimally modifying the original actions to render the planning task unsolvable. We empirically evaluate the performance of $allmin$ in shielding planning tasks of increasing size, showing how it can effectively shield the system by turning the planning task unsolvable.

Text-to-SQL is the task of translating natural language queries into executable SQL for a given database, enabling non-expert users to access structured data without writing SQL manually. Despite rapid advances driven by large language models (LLMs), existing approaches still struggle with complex queries in real-world settings, where database schemas are large and questions require multi-step reasoning over many interrelated tables. In such cases, providing the full schema often exceeds the context window, while one-shot generation frequently produces non-executable SQL due to syntax errors and incorrect schema linking. To address these challenges, we introduce AV-SQL, a framework that decomposes complex Text-to-SQL into a pipeline of specialized LLM agents. Central to AV-SQL is the concept of agentic views: agent-generated Common Table Expressions (CTEs) that encapsulate intermediate query logic and filter relevant schema elements from large schemas. AV-SQL operates in three stages: (1) a rewriter agent compresses and clarifies the input query; (2) a view generator agent processes schema chunks to produce agentic views; and (3) a planner, generator, and revisor agent collaboratively compose these views into the final SQL query. Extensive experiments show that AV-SQL achieves 70.38% execution accuracy on the challenging Spider 2.0 benchmark, outperforming state-of-the-art baselines, while remaining competitive on standard datasets with 85.59% on Spider, 72.16% on BIRD and 63.78% on KaggleDBQA. Our source code is available at https://github.com/pminhtam/AV-SQL.

Robotic systems lack a principled abstraction for organizing intelligence, capabilities, and execution in a unified manner. Existing approaches either couple skills within monolithic architectures or decompose functionality into loosely coordinated modules or multiple agents, often without a coherent model of identity and control authority. We argue that a robot should be modeled as a single persistent intelligent subject whose capabilities are extended through installable packages. We formalize this view as AEROS (Agent Execution Runtime Operating System), in which each robot corresponds to one persistent agent and capabilities are provided through Embodied Capability Modules (ECMs). Each ECM encapsulates executable skills, models, and tools, while execution constraints and safety guarantees are enforced by a policy-separated runtime. This separation enables modular extensibility, composable capability execution, and consistent system-level safety. We evaluate a reference implementation in PyBullet simulation with a Franka Panda 7-DOF manipulator across eight experiments covering re-planning, failure recovery, policy enforcement, baseline comparison, cross-task generality, ECM hot-swapping, ablation, and failure boundary analysis. Over 100 randomized trials per condition, AEROS achieves 100% task success across three tasks versus baselines (BehaviorTree.CPP-style and ProgPrompt-style at 92--93%, flat pipeline at 67--73%), the policy layer blocks all invalid actions with zero false acceptances, runtime benefits generalize across tasks without task-specific tuning, and ECMs load at runtime with 100% post-swap success.

Recently, LLM-based agents have become increasingly popular across many applications, including complex sequential decision-making problems. However, they inherit the tendency of LLMs to hallucinate, leading to incorrect decisions. In sequential settings, even a single mistake can irreversibly degrade the trajectory, making hallucinations an even bigger problem. Although larger LLMs hallucinate less, they incur a significantly higher per-token cost. In this paper, we address this tradeoff by proposing ReDAct (Reason-Defer-Act). In ReDAct, an agent is equipped with two LLMs: a small, cheap model used by default, and a large, more reliable but expensive model. When the predictive uncertainty of the small model exceeds a calibrated threshold, the decision is deferred to the large model. We evaluate our approach in text-based embodied environments such as ALFWorld and MiniGrid and show that deferring only about 15% of decisions to the large model can match the quality of using it exclusively, while significantly reducing inference costs.

Mixture-of-experts (MoE) language models are often expected to offer better quality-efficiency tradeoffs than dense models because only a subset of parameters is activated per token, but the practical value of that advantage depends on end-to-end behavior under realistic inference constraints. We present a controlled empirical benchmark of seven recent reasoning-oriented instruction-tuned models spanning dense and MoE designs, namely Gemma-4-E2B, Gemma-4-E4B, Gemma-4-26B-A4B, Phi-4-mini-reasoning, Phi-4-reasoning, Qwen3-8B, and Qwen3-30B-A3B, evaluated on four benchmarks -- ARC-Challenge, GSM8K, Math Level 1-3, and TruthfulQA MC1 -- under three prompting strategies: zero-shot, chain-of-thought, and few-shot chain-of-thought. The study covers 8,400 total model-dataset-prompt evaluations and records accuracy, latency, peak GPU memory usage (VRAM), and an approximate floating-point operations (FLOPs)-per-token proxy. Across the weighted multi-task summary, Gemma-4-E4B with few-shot chain-of-thought achieved the best overall result, reaching weighted accuracy 0.675 with mean VRAM 14.9 GB, while Gemma-4-26B-A4B was close in accuracy at 0.663 but substantially more memory intensive at 48.1 GB. At the task level, Gemma models dominated ARC and Math, Phi models were strongest on TruthfulQA, and GSM8K showed the largest prompt sensitivity, including a sharp drop for Phi-4-reasoning from 0.67 under chain-of-thought to 0.11 under few-shot chain-of-thought. These results show that sparse activation alone does not guarantee the best practical operating point: observed accuracy-efficiency tradeoffs depend jointly on architecture, prompting protocol, and task composition. We release a reproducible benchmark pipeline, aggregated results, and paired statistical analyses to support deployment-oriented evaluation of reasoning LLMs under real resource constraints.

We present KITE, a training-free, keyframe-anchored, layout-grounded front-end that converts long robot-execution videos into compact, interpretable tokenized evidence for vision-language models (VLMs). KITE distills each trajectory into a small set of motion-salient keyframes with open-vocabulary detections and pairs each keyframe with a schematic bird's-eye-view (BEV) representation that encodes relative object layout, axes, timestamps, and detection confidence. These visual cues are serialized with robot-profile and scene-context tokens into a unified prompt, allowing the same front-end to support failure detection, identification, localization, explanation, and correction with an off-the-shelf VLM. On the RoboFAC benchmark, KITE with Qwen2.5-VL substantially improves over vanilla Qwen2.5-VL in the training-free setting, with especially large gains on simulation failure detection, identification, and localization, while remaining competitive with a RoboFAC-tuned baseline. A small QLoRA fine-tune further improves explanation and correction quality. We also report qualitative results on real dual-arm robots, demonstrating the practical applicability of KITE as a structured and interpretable front-end for robot failure analysis. Code and models are released on our project page: https://m80hz.github.io/kite/

Sparse Mixture-of-Experts (MoE) architectures are increasingly popular for frontier large language models (LLM) but they introduce training challenges due to routing complexity. Fully leveraging parameters of an MoE model requires all experts to be well-trained and to specialize in non-redundant ways. Assessing this, however, is complicated due to lack of established metrics and, importantly, many routing techniques exhibit similar performance at smaller sizes, which is often not reflective of their behavior at large scale. To address this challenge, we propose the MoE Routing Testbed, a setup that gives clearer visibility into routing dynamics at small scale while using realistic data. The testbed pairs a data mix with clearly distinguishable domains with a reference router that prescribes ideal routing based on these domains, providing a well-defined upper bound for comparison. This enables quantifiable measurement of expert specialization. To demonstrate the value of the testbed, we compare various MoE routing approaches and show that balancing scope is the crucial factor that allows specialization while maintaining high expert utilization. We confirm that this observation generalizes to models 35x larger.

Strategic interaction in adversarial domains such as law, diplomacy, and negotiation is mediated by language, yet most game-theoretic models abstract away the mechanisms of persuasion that operate through discourse. We present the Strategic Courtroom Framework, a multi-agent simulation environment in which prosecution and defense teams composed of trait-conditioned Large Language Model (LLM) agents engage in iterative, round-based legal argumentation. Agents are instantiated using nine interpretable traits organized into four archetypes, enabling systematic control over rhetorical style and strategic orientation. We evaluate the framework across 10 synthetic legal cases and 84 three-trait team configurations, totaling over 7{,}000 simulated trials using DeepSeek-R1 and Gemini~2.5~Pro. Our results show that heterogeneous teams with complementary traits consistently outperform homogeneous configurations, that moderate interaction depth yields more stable verdicts, and that certain traits (notably quantitative and charismatic) contribute disproportionately to persuasive success. We further introduce a reinforcement-learning-based Trait Orchestrator that dynamically generates defense traits conditioned on the case and opposing team, discovering strategies that outperform static, human-designed trait combinations. Together, these findings demonstrate how language can be treated as a first-class strategic action space and provide a foundation for building autonomous agents capable of adaptive persuasion in multi-agent environments.

Nonstationarity is ubiquitous in practical classification settings, leading deployed models to perform poorly even when they generalize well to holdout sets available at training time. We address this by reframing nonstationary classification as time series prediction: rather than predicting from the current input alone, we condition the classifier on a sequence of historical labeled examples that extends beyond the training cutoff. To scale to large sequences, we introduce a learned discrete retrieval mechanism that samples relevant historical examples via input-dependent queries, trained end-to-end with the classifier using a score-based gradient estimator. This enables the full corpus of historical data to remain on an arbitrary filesystem during training and deployment. Experiments on synthetic benchmarks and Amazon Reviews '23 (electronics category) show improved robustness to distribution shift compared to standard classifiers, with VRAM scaling predictably as the length of the historical data sequence increases.

This article presents a novel and comprehensive approach for analyzing bending behavior of the tapered perforated beam under an exponential load. The governing differential equation includes important factors like filling ratio ($α$), number of rows of holes ($N$), tapering parameters ($φ$ and $ψ$), and exponential loading parameter ($γ$), providing a realistic and flexible representation of perforated beam configuration. Main goal of this work is to see how well the Domain mapped physics-informed Functional link Theory of Functional Connection (DFL-TFC) method analyses bending response of perforated beam with square holes under exponential loading. For comparison purposes, a corresponding PINN-based formulation is developed. Outcomes clearly show that the proposed DFL-TFC framework gives better results, including faster convergence, reduced computational cost, and improved solution accuracy when compared to the PINN approach. These findings highlight effectiveness and potential of DFL-TFC method for solving complex engineering problems governed by differential equations. Within this framework, hidden layer is replaced by a functional expansion block that enriches input representation via orthogonal polynomial basis functions, and the domain of DE mapped to corresponding domain of orthogonal polynomials. A Constrained Expression (CE), constructed through the Theory of Functional Connections (TFC) using boundary conditions, ensures that constraints are exactly satisfied. In CE, free function is represented using a Functional Link Neural Network (FLNN), which learns to solve resulting unconstrained optimization problem. The obtained results are further validated through the Galerkin and PINN solutions.

Autoregressive (AR) language models generate text one token at a time, even when consecutive tokens are highly predictable given earlier context. We introduce MARS (Mask AutoRegreSsion), a lightweight fine-tuning method that teaches an instruction-tuned AR model to predict multiple tokens per forward pass. MARS adds no architectural modifications, no extra parameters, and produces a single model that can still be called exactly like the original AR model with no performance degradation. Unlike speculative decoding, which maintains a separate draft model alongside the target, or multi-head approaches such as Medusa, which attach additional prediction heads, MARS requires only continued training on existing instruction data. When generating one token per forward pass, MARS matches or exceeds the AR baseline on six standard benchmarks. When allowed to accept multiple tokens per step, it maintains baseline-level accuracy while achieving 1.5-1.7x throughput. We further develop a block-level KV caching strategy for batch inference, achieving up to 1.71x wall-clock speedup over AR with KV cache on Qwen2.5-7B. Finally, MARS supports real-time speed adjustment via confidence thresholding: under high request load, the serving system can increase throughput on the fly without swapping models or restarting, providing a practical latency-quality knob for deployment.

Neural networks deliver impressive predictive performance across a variety of tasks, but they are often opaque in their decision-making processes. Despite a growing interest in mechanistic interpretability, tools for systematically exploring the representations learned by neural networks in general, and tabular foundation models in particular, remain limited. In this work, we introduce ConceptTracer, an interactive application for analyzing neural representations through the lens of human-interpretable concepts. ConceptTracer integrates two information-theoretic measures that quantify concept saliency and selectivity, enabling researchers and practitioners to identify neurons that respond strongly to individual concepts. We demonstrate the utility of ConceptTracer on representations learned by TabPFN and show that our approach facilitates the discovery of interpretable neurons. Together, these capabilities provide a practical framework for investigating how neural networks like TabPFN encode concept-level information. ConceptTracer is available at https://github.com/ml-lab-htw/concept-tracer.

AI assistants that interact with users over time need to interpret the user's current emotional state in order to respond appropriately and personally. However, this capability remains insufficiently evaluated. Existing emotion datasets mainly assess local or instantaneous affect, while long-term memory benchmarks focus largely on factual recall, temporal consistency, or knowledge updating. As a result, current resources provide limited support for testing whether a model can use remembered interaction history to interpret a user's present affective state. We introduce A-MBER, an Affective Memory Benchmark for Emotion Recognition, to evaluate this capability. A-MBER focuses on present affective interpretation grounded in remembered multi-session interaction history. Given an interaction trajectory and a designated anchor turn, a model must infer the user's current affective state, identify historically relevant evidence, and justify its interpretation in a grounded way. The benchmark is constructed through a staged pipeline with explicit intermediate representations, including long-horizon planning, conversation generation, annotation, question construction, and final packaging. It supports judgment, retrieval, and explanation tasks, together with robustness settings such as modality degradation and insufficient-evidence conditions. Experiments compare local-context, long-context, retrieved-memory, structured-memory, and gold-evidence conditions within a unified framework. Results show that A-MBER is especially discriminative on the subsets it is designed to stress, including long-range implicit affect, high-dependency memory levels, trajectory-based reasoning, and adversarial settings. These findings suggest that memory supports affective interpretation not simply by providing more history, but by enabling more selective, grounded, and context-sensitive use of past interaction

A key challenge in scaling up Reinforcement Learning is generalizing learned behaviour. Without the ability to carry forward acquired knowledge an agent is doomed to learn each task from scratch. In this paper we develop a new formalism for transfer by virtue of state abstraction. Based on task-independent, compact observations (outcomes) of the environment, we introduce Outcome-Predictive State Representations (OPSRs), agent-centered and task-independent abstractions that are made up of predictions of outcomes. We show formally and empirically that they have the potential for optimal but limited transfer, then overcome this trade-off by introducing OPSR-based skills, i.e. abstract actions (based on options) that can be reused between tasks as a result of state abstraction. In a series of empirical studies, we learn OPSR-based skills from demonstrations and show how they speed up learning considerably in entirely new and unseen tasks without any pre-processing. We believe that the framework introduced in this work is a promising step towards transfer in RL in general, and towards transfer through combining state and action abstraction specifically.

In this article, we seek to answer the following question: could data duplication be useful in Natural Language Processing (NLP) for languages with limited computational resources? In this type of languages (or $π$-languages), corpora available for training Large Language Models are virtually non-existent. In particular, we will study the impact of corpora expansion in Nawatl, an agglutinative and polysynthetic $π$-language spoken by over 2 million people, with a large number of dialectal varieties. The aim is to expand the new $π$-yalli corpus, which contains a limited number of Nawatl texts, by duplicating it in a controlled way. In our experiments, we will use the incremental duplication technique. The aim is to learn embeddings that are well-suited to NLP tasks. Thus, static embeddings were trained and evaluated in a sentence-level semantic similarity task. Our results show a moderate improvement in performance when using incremental duplication compared to the results obtained using only the corpus without expansion. Furthermore, to our knowledge, this technique has not yet been used in the literature.

Designing quantum neural networks (QNNs) that are both accurate and deployable on NISQ hardware is challenging. Handcrafted ansatze must balance expressivity, trainability, and resource use, while limited qubits often necessitate circuit cutting. Existing quantum architecture search methods primarily optimize accuracy while only heuristically controlling quantum and mostly ignore the exponential overhead of circuit cutting. We introduce QNAS, a neural architecture search framework that unifies hardware aware evaluation, multi objective optimization, and cutting overhead awareness for hybrid quantum classical neural networks (HQNNs). QNAS trains a shared parameter SuperCircuit and uses NSGA-II to optimize three objectives jointly: (i) validation error, (ii) a runtime cost proxy measuring wall clock evaluation time, and (iii) the estimated number of subcircuits under a target qubit budget. QNAS evaluates candidate HQNNs under a few epochs of training and discovers clear Pareto fronts that reveal tradeoffs between accuracy, efficiency, and cutting overhead. Across MNIST, Fashion-MNIST, and Iris benchmarks, we observe that embedding type and CNOT mode selection significantly impact both accuracy and efficiency, with angle-y embedding and sparse entangling patterns outperforming other configurations on image datasets, and amplitude embedding excelling on tabular data (Iris). On MNIST, the best architecture achieves 97.16% test accuracy with a compact 8 qubit, 2 layer circuit; on the more challenging Fashion-MNIST, 87.38% with a 5 qubit, 2 layer circuit; and on Iris, 100% validation accuracy with a 4 qubit, 2 layer circuit. QNAS surfaces these design insights automatically during search, guiding practitioners toward architectures that balance accuracy, resource efficiency, and practical deployability on current hardware.

Retrieval-Augmented Generation (RAG) mitigates the hallucination problem of Large Language Models (LLMs) by incorporating external knowledge. Recursive summarization constructs a hierarchical summary tree by clustering text chunks, integrating information from multiple parts of a document to provide evidence for abstractive questions involving multi-step reasoning. However, summary trees often contain a large number of redundant summary nodes, which not only increase construction time but may also negatively impact question answering. Moreover, recursive summarization is not suitable for all types of questions. We introduce DTCRS, a method that dynamically generates summary trees based on document structure and query semantics. DTCRS determines whether a summary tree is necessary by analyzing the question type. It then decomposes the question and uses the embeddings of sub-questions as initial cluster centers, reducing redundant summaries while improving the relevance between summaries and the question. Our approach significantly reduces summary tree construction time and achieves substantial improvements across three QA tasks. Additionally, we investigate the applicability of recursive summarization to different question types, providing valuable insights for future research.

Ensuring fairness in machine learning predictions is a critical challenge, especially when models are deployed in sensitive domains such as credit scoring, healthcare, and criminal justice. While many fairness interventions rely on data preprocessing or algorithmic constraints during training, these approaches often require full control over the model architecture and access to protected attribute information, which may not be feasible in real-world systems. In this paper, we propose Counterfactual Averaging for Fair Predictions (CAFP), a model-agnostic post-processing method that mitigates unfair influence from protected attributes without retraining or modifying the original classifier. CAFP operates by generating counterfactual versions of each input in which the sensitive attribute is flipped, and then averaging the model's predictions across factual and counterfactual instances. We provide a theoretical analysis of CAFP, showing that it eliminates direct dependence on the protected attribute, reduces mutual information between predictions and sensitive attributes, and provably bounds the distortion introduced relative to the original model. Under mild assumptions, we further show that CAFP achieves perfect demographic parity and reduces the equalized odds gap by at least half the average counterfactual bias.

Autonomous AI agents are beginning to operate across organizational boundaries on the open internet -- discovering, transacting with, and delegating to agents owned by other parties without centralized oversight. When agents from different human principals collaborate at scale, the collective becomes opaque: no single human can observe, audit, or govern the emergent behavior. We term this the Logic Monopoly -- the agent society's unchecked monopoly over the entire logic chain from planning through execution to evaluation. We propose the Separation of Power (SoP) model, a constitutional governance architecture deployed on public blockchain that breaks this monopoly through three structural separations: agents legislate operational rules as smart contracts, deterministic software executes within those contracts, and humans adjudicate through a complete ownership chain binding every agent to a responsible principal. In this architecture, smart contracts are the law itself -- the actual legislative output that agents produce and that governs their behavior. We instantiate SoP in AgentCity on an EVM-compatible layer-2 blockchain (L2) with a three-tier contract hierarchy (foundational, meta, and operational). The core thesis is alignment-through-accountability: if each agent is aligned with its human owner through the accountability chain, then the collective converges on behavior aligned with human intent -- without top-down rules. A pre-registered experiment evaluates this thesis in a commons production economy -- where agents share a finite resource pool and collaboratively produce value -- at 50-1,000 agent scale.

Pragmatic inference is inherently graded. Different lexical items give rise to pragmatic enrichment to different degrees. Scalar implicature exemplifies this property through scalar diversity, where implicature strength varies across scalar items. However, evaluations of pragmatic inference in large language models (LLMs) often rely on prompt-based manipulations. Beyond prompt-level effects, this study introduces Continuous Interpretive Steering (CIS), a method that probes graded pragmatic interpretation by treating activation-level steering strength as a continuous experimental variable. To support this analysis, this study introduces a new dataset, GraSD, which encodes graded scalar diversity. Experiments on four LLMs show that uniform activation steering increases pragmatic interpretations globally but collapses item-level variation, whereas graded activation steering yields differentiated interpretive shifts aligned with scalar diversity grades. It indicates that graded sensitivity is encoded in the representation space and can be systematically recovered through controlled intervention. Together, CIS and GraSD provide a principled framework for evaluating graded pragmatic sensitivity in LLMs.

Large language models (LLMs) has been widely used for automated negotiation, but their high computational cost and privacy risks limit deployment in privacy-sensitive, on-device settings such as mobile assistants or rescue robots. Small language models (SLMs) offer a viable alternative, yet struggle with the complex emotional dynamics of high-stakes negotiation. We introduces EmoMAS, a Bayesian multi-agent framework that transforms emotional decision-making from reactive to strategic. EmoMAS leverages a Bayesian orchestrator to coordinate three specialized agents: game-theoretic, reinforcement learning, and psychological coherence models. The system fuses their real-time insights to optimize emotional state transitions while continuously updating agent reliability based on negotiation feedback. This mixture-of-agents architecture enables online strategy learning without pre-training. We further introduce four high-stakes, edge-deployable negotiation benchmarks across debt, healthcare, emergency response, and educational domains. Through extensive agent-to-agent simulations across all benchmarks, both SLMs and LLMs equipped with EmoMAS consistently surpass all baseline models in negotiation performance while balancing ethical behavior. These results show that strategic emotional intelligence is also the key driver of negotiation success. By treating emotional expression as a strategic variable within a Bayesian multi-agent optimization framework, EmoMAS establishes a new paradigm for effective, private, and adaptive negotiation AI suitable for high-stakes edge deployment.

Retrieval shapes how language models access and ground knowledge in retrieval-augmented generation (RAG). In historical research, the target is often not an arbitrary relevant passage, but the exact record for a specific regnal month, where temporal consistency matters as much as topical relevance. This is especially challenging for Classical Chinese annals, where time is expressed through terse, implicit, non-Gregorian reign phrases that must be interpreted from surrounding context, so semantically plausible evidence can still be temporally invalid. We introduce \textbf{ChunQiuTR}, a time-keyed retrieval benchmark built from the \textit{Spring and Autumn Annals} and its exegetical tradition. ChunQiuTR organizes records by month-level reign keys and includes chrono-near confounders that mirror realistic retrieval failures. We further propose \textbf{CTD} (Calendrical Temporal Dual-encoder), a time-aware dual-encoder that combines Fourier-based absolute calendrical context with relative offset biasing. Experiments show consistent gains over strong semantic dual-encoder baselines under time-keyed evaluation, supporting retrieval-time temporal consistency as a key prerequisite for faithful downstream historical RAG. Our code and datasets are available at \href{https://github.com/xbdxwyh/ChunQiuTR}{\texttt{github.com/xbdxwyh/ChunQiuTR}}.

LLM-as-a-judge has become the de facto approach for evaluating LLM outputs. However, judges are known to exhibit self-preference bias (SPB): they tend to favor outputs produced by themselves or by models from their own family. This skews evaluations and, thus, hinders model development, especially in settings of recursive self-improvement. We present the first study of SPB in rubric-based evaluation, an increasingly popular benchmarking paradigm where judges issue binary verdicts on individual evaluation criteria, instead of assigning holistic scores or rankings. Using IFEval, a benchmark with programmatically verifiable rubrics, we show that SPB persists even when evaluation criteria are entirely objective: among rubrics where generators fail, judges can be up to 50\% more likely to incorrectly mark them as satisfied when the output is their own. We also find that, similarly to other evaluation paradigms, ensembling multiple judges helps mitigate SPB, but without fully eliminating it. On HealthBench, a medical chat benchmark with subjective rubrics, we observe that SPB skews model scores by up to 10 points, a potentially decisive margin when ranking frontier models. We analyze the factors that drive SPB in this setting, finding that negative rubrics, extreme rubric lengths, and subjective topics like emergency referrals are particularly susceptible.

Existing Graphical User Interface (GUI) reasoning tasks remain challenging, particularly in UI understanding. Current methods typically rely on direct screen-based decision-making, which lacks interpretability and overlooks a comprehensive understanding of UI elements, ultimately leading to task failure. To enhance the understanding and interaction with UIs, we propose an innovative GUI reasoning paradigm called UI-in-the-Loop (UILoop). Our approach treats the GUI reasoning task as a cyclic Screen-UI elements-Action process. By enabling Multimodal Large Language Models (MLLMs) to explicitly learn the localization, semantic functions, and practical usage of key UI elements, UILoop achieves precise element discovery and performs interpretable reasoning. Furthermore, we introduce a more challenging UI Comprehension task centered on UI elements with three evaluation metrics. Correspondingly, we contribute a benchmark of 26K samples (UI Comprehension-Bench) to comprehensively evaluate existing methods' mastery of UI elements. Extensive experiments demonstrate that UILoop achieves state-of-the-art UI understanding performance while yielding superior results in GUI reasoning tasks.

Psychological stress is clinically relevant in cardio-oncology, yet it is typically assessed only through patient-reported outcome measures (PROMs) and is rarely integrated into continuous cardiotoxicity surveillance. We estimate perceived stress in an elderly, multicenter breast cancer cohort (CARDIOCARE) using multimodal wearable data from a smartwatch (physical activity and sleep) and a chest-worn ECG sensor. Wearable streams are transformed into heterogeneous visual representations, yielding a weakly supervised setting in which a single Perceived Stress Scale (PSS) score corresponds to many unlabeled windows. A lightweight pretrained mixture-of-experts backbone (Tiny-BioMoE) embeds each representation into 192-dimensional vectors, which are aggregated via attention-based multiple instance learning (MIL) to predict PSS at month 3 (M3) and month 6 (M6). Under leave-one-subject-out (LOSO) evaluation, predictions showed moderate agreement with questionnaire scores (M3: R^2=0.24, Pearson r=0.42, Spearman rho=0.48; M6: R^2=0.28, Pearson r=0.49, Spearman rho=0.52), with global RMSE/MAE of 6.62/6.07 at M3 and 6.13/5.54 at M6.

We present the first generative approach to photomosaic creation. Traditional photomosaic methods rely on a large number of tile images and color-based matching, which limits both diversity and structural consistency. Our generative photomosaic framework synthesizes tile images using diffusion-based generation conditioned on reference images. A low-frequency conditioned diffusion mechanism aligns global structure while preserving prompt-driven details. This generative formulation enables photomosaic composition that is both semantically expressive and structurally coherent, effectively overcoming the fundamental limitations of matching-based approaches. By leveraging few-shot personalized diffusion, our model is able to produce user-specific or stylistically consistent tiles without requiring an extensive collection of images.

Palmprint recognition is deployed in security-critical applications, including access control and palm-based payment, due to its contactless acquisition and highly discriminative ridge-and-crease textures. However, the robustness of deep palmprint recognition systems against physically realizable attacks remains insufficiently understood. Existing studies are largely confined to the digital setting and do not adequately account for the texture-dominant nature of palmprint recognition or the distortions introduced during physical acquisition. To address this gap, we propose CAAP, a capture-aware adversarial patch framework for palmprint recognition. CAAP learns a universal patch that can be reused across inputs while remaining effective under realistic acquisition variation. To match the structural characteristics of palmprints, the framework adopts a cross-shaped patch topology, which enlarges spatial coverage under a fixed pixel budget and more effectively disrupts long-range texture continuity. CAAP further integrates three modules: ASIT for input-conditioned patch rendering, RaS for stochastic capture-aware simulation, and MS-DIFE for feature-level identity-disruptive guidance. We evaluate CAAP on the Tongji, IITD, and AISEC datasets against generic CNN backbones and palmprint-specific recognition models. Experiments show that CAAP achieves strong untargeted and targeted attack performance with favorable cross-model and cross-dataset transferability. The results further show that, although adversarial training can partially reduce the attack success rate, substantial residual vulnerability remains. These findings indicate that deep palmprint recognition systems remain vulnerable to physically realizable, capture-aware adversarial patch attacks, underscoring the need for more effective defenses in practice. Code available at https://github.com/ryliu68/CAAP.

Frailty and functional decline strongly influence treatment tolerance and outcomes in older patients with cancer, yet assessment is typically limited to infrequent clinic visits. We propose a multimodal wearable framework to estimate frailty-related functional change between visits in elderly breast cancer patients enrolled in the multicenter CARDIOCARE study. Free-living smartwatch physical activity and sleep features are combined with ECG-derived heart rate variability (HRV) features from a chest strap and organized into patient-horizon bags aligned to month 3 (M3) and month 6 (M6) follow-ups. Our innovation is an attention-based multiple instance learning (MIL) formulation that fuses irregular, multimodal wearable instances under real-world missingness and weak supervision. An attention-based MIL model with modality-specific multilayer perceptron (MLP) encoders with embedding dimension 128 aggregates variable-length and partially missing longitudinal instances to predict discretized change-from-baseline classes (worsened, stable, improved) for FACIT-F and handgrip strength. Under subject-independent leave-one-subject-out (LOSO) evaluation, the full multimodal model achieved balanced accuracy/F1 of 0.68 +/- 0.08/0.67 +/- 0.09 at M3 and 0.70 +/- 0.10/0.69 +/- 0.08 at M6 for handgrip, and 0.59 +/- 0.04/0.58 +/- 0.06 at M3 and 0.64 +/- 0.05/0.63 +/- 0.07 at M6 for FACIT-F. Ablation results indicated that smartwatch activity and sleep provide the strongest predictive information for frailty-related functional changes, while HRV contributes complementary information when fused with smartwatch streams.

Reliable radar pulse classification is essential in Electromagnetic Warfare for situational awareness and decision support. Deep Neural Networks have shown strong performance in radar pulse and RF emitter recognition; however, on their own they struggle to efficiently learn new pulses and lack mechanisms for expressing predictive confidence. This paper integrates Uncertainty Quantification with Lifelong Learning to address both challenges. The proposed approach is an Evidential Lifelong Classifier (ELC), which models epistemic uncertainty using evidence theory. ELC is evaluated against a Bayesian Lifelong Classifier (BLC), which quantifies uncertainty through Shannon entropy. Both integrate Learn-Prune-Share to enable continual learning of new pulses and uncertainty-based selective prediction to reject unreliable predictions. ELC and BLC are evaluated on 2 synthetic radar and 3 RF fingerprinting datasets. Selective prediction based on evidential uncertainty improves recall by up to 46% at -20 dB SNR on synthetic radar pulse datasets, highlighting its effectiveness at identifying unreliable predictions in low-SNR conditions compared to BLC. These findings demonstrate that evidential uncertainty offers a strong correlation between confidence and correctness, improving the trustworthiness of ELC by allowing it to express ignorance.

Modern recommendation models have increased to trillions of parameters. As cluster scales expand to O(1k), distributed training bottlenecks shift from computation and memory to data movement, especially lookup and communication latency associated with embeddings. Existing solutions either optimize only one bottleneck or improve throughput by sacrificing training consistency. This paper presents NestPipe, a large-scale decentralized embedding training framework that tackles both bottlenecks while preserving synchronous training semantics. NestPipe exploits two hierarchical sparse parallelism opportunities through nested pipelining. At the inter-batch level, Dual-Buffer Pipelining (DBP) constructs a staleness-free five-stage pipeline through dual-buffer synchronization, mitigating lookup bottlenecks without embedding staleness. At the intra-batch level, we identify the embedding freezing phenomenon, which inspires Frozen-Window Pipelining (FWP) to overlap All2All communication with dense computation via coordinated stream scheduling and key-centric sample clustering. Experiments on production GPU and NPU clusters with 1,536 workers demonstrate that NestPipe achieves up to 3.06x speedup and 94.07% scaling efficiency.

Automated test case generation from natural language requirements remains a challenging problem in software engineering due to the ambiguity of requirements and the need to produce structured, executable test artifacts. Recent advances in LLMs have shown promise in addressing this task; however, their effectiveness depends on task-specific adaptation and efficient fine-tuning strategies. In this paper, we present a comprehensive empirical study on the use of parameter-efficient fine-tuning, specifically LoRA, for requirement-based test case generation. We evaluate multiple LLM families, including open-source and proprietary models, under a unified experimental pipeline. The study systematically explores the impact of key LoRA hyperparameters, including rank, scaling factor, and dropout, on downstream performance. We propose an automated evaluation framework based on GPT-4o, which assesses generated test cases across nine quality dimensions. Experimental results demonstrate that LoRA-based fine-tuning significantly improves the performance of all open-source models, with Ministral-8B achieving the best results among them. Furthermore, we show that a fine-tuned 8B open-source model can achieve performance comparable to pre-fine-tuned GPT-4.1 models, highlighting the effectiveness of parameter-efficient adaptation. While GPT-4.1 models achieve the highest overall performance, the performance gap between proprietary and open-source models is substantially reduced after fine-tuning. These findings provide important insights into model selection, fine-tuning strategies, and evaluation methods for automated test generation. In particular, they demonstrate that cost-efficient, locally deployable open-source models can serve as viable alternatives to proprietary systems when combined with well-designed fine-tuning approaches.

Ensuring ciphertext indistinguishability is fundamental to cryptographic security, but empirically validating this property in real implementations and hybrid settings presents practical challenges. The transition to post-quantum cryptography (PQC), with its hybrid constructions combining classical and quantum-resistant primitives, makes empirical validation approaches increasingly valuable. By modeling IND-CPA games as binary classification tasks and training on labeled ciphertext data with BCE loss, we study deep neural network (DNN) distinguishers for ciphertext indistinguishability. We apply this methodology to PQC KEMs. We specifically test the public-key encryption (PKE) schemes used to construct examples such as ML-KEM, BIKE, and HQC. Moreover, a novel extension of this DNN modeling for empirical distinguishability testing of hybrid KEMs is presented. We implement and test this on combinations of PQC KEMs with plain RSA, RSA-OAEP, and plaintext. Finally, methodological generality is illustrated by applying the DNN IND-CPA classification framework to cascade symmetric encryption, where we test combinations of AES-CTR, AES-CBC, AES-ECB, ChaCha20, and DES-ECB. In our experiments on PQC algorithms, KEM combiners, and cascade encryption, no algorithm or combination of algorithms demonstrates a significant advantage (two-sided binomial test, significance level $α= 0.01$), consistent with theoretical guarantees that hybrids including at least one IND-CPA-secure component preserve indistinguishability, and with the absence of exploitable patterns under the considered DNN adversary model. These illustrate the potential of using deep learning as an adaptive, practical, and versatile empirical estimator for indistinguishability in more general IND-CPA settings, allowing data-driven validation of implementations and compositions and complementing the analytical security analysis.

Most neural solvers for the Traveling Salesperson Problem (TSP) are trained to output a single solution, even though practitioners rarely stop there: at test time, they routinely spend extra compute on sampling or post-hoc search. This raises a natural question: can the search procedure itself be learned? Neural improvement methods take this perspective by learning a policy that applies local modifications to a candidate solution, accumulating gains over an improvement trajectory. Yet learned improvement for TSP remains comparatively immature, with existing methods still falling short of robust, scalable performance. We argue that a key reason is design mismatch: many approaches reuse state representations, architectural choices, and training recipes inherited from single-solution methods, rather than being built around the mechanics of local search. This mismatch motivates NICO-TSP (Neural Improvement for Combinatorial Optimization): a 2-opt improvement framework for TSP. NICO-TSP represents the current tour with exactly $n$ edge tokens aligned with the neighborhood operator, scores 2-opt moves directly without tour positional encodings, and trains via a two-stage procedure: imitation learning to short-horizon optimal trajectories, followed by critic-free group-based reinforcement learning over longer rollouts. Under compute-matched evaluations that measure improvement as a function of both search steps and wall-clock time, NICO-TSP delivers consistently stronger and markedly more step-efficient improvement than prior learned and heuristic search baselines, generalizes far more reliably to larger out-of-distribution instances, and serves both as a competitive replacement for classical local search and as a powerful test-time refinement module for constructive solvers.

Detecting user interface (UI) controls from software screenshots is a critical task for automated testing, accessibility, and software analytics, yet it remains challenging due to visual ambiguities, design variability, and the lack of contextual cues in pixel-only approaches. In this paper, we introduce a novel multi-modal extension of YOLOv5 that integrates GPT-generated textual descriptions of UI images into the detection pipeline through cross-attention modules. By aligning visual features with semantic information derived from text embeddings, our model enables more robust and context-aware UI control detection. We evaluate the proposed framework on a large dataset of over 16,000 annotated UI screenshots spanning 23 control classes. Extensive experiments compare three fusion strategies, i.e. element-wise addition, weighted sum, and convolutional fusion, demonstrating consistent improvements over the baseline YOLOv5 model. Among these, convolutional fusion achieved the strongest performance, with significant gains in detecting semantically complex or visually ambiguous classes. These results establish that combining visual and textual modalities can substantially enhance UI element detection, particularly in edge cases where visual information alone is insufficient. Our findings open promising opportunities for more reliable and intelligent tools in software testing, accessibility support, and UI analytics, setting the stage for future research on efficient, robust, and generalizable multi-modal detection systems.

We study the continuous-time structure of the difference-of-convex algorithm (DCA) for smooth DC decompositions with a strongly convex component. In dual coordinates, classical DCA is exactly the full-step explicit Euler discretization of a nonlinear autonomous system. This viewpoint motivates a damped DCA scheme, which is also a Bregman-regularized DCA variant, and whose vanishing-step limit yields a Hessian-Riemannian gradient flow generated by the convex part of the decomposition. For the damped scheme we prove monotone descent, asymptotic criticality, Kurdyka-Lojasiewicz convergence under boundedness, and a global linear rate under a metric DC-PL inequality. For the limiting flow we establish an exact energy identity, asymptotic criticality of bounded trajectories, explicit global rates under metric relative error bounds, finite-length and single-point convergence under a Kurdyka-Lojasiewicz hypothesis, and local exponential convergence near nondegenerate local minima. The analysis also reveals a global-local tradeoff: the half-relaxed scheme gives the best provable global guarantee in our framework, while the full-step scheme is locally fastest near a nondegenerate minimum. Finally, we show that different DC decompositions of the same objective induce different continuous dynamics through the metric generated by the convex component, providing a geometric criterion for decomposition quality and linking DCA with Bregman geometry.

Reinforcement-Learning-based post-training has recently emerged as a promising paradigm for aligning text-to-image diffusion models with human preferences. In recent studies, increasing the rollout group size yields pronounced performance improvements, indicating substantial room for further alignment gains. However, scaling rollouts on large-scale foundational diffusion models (e.g., FLUX.1-12B) imposes a heavy computational burden. To alleviate this bottleneck, we explore the integration of FP4 quantization into Diffusion RL rollouts. Yet, we identify that naive quantized pipelines inherently introduce risks of performance degradation. To overcome this dilemma between efficiency and training integrity, we propose Sol-RL (Speed-of-light RL), a novel FP4-empowered Two-stage Reinforcement Learning framework. First, we utilize high-throughput NVFP4 rollouts to generate a massive candidate pool and extract a highly contrastive subset. Second, we regenerate these selected samples in BF16 precision and optimize the policy exclusively on them. By decoupling candidate exploration from policy optimization, Sol-RL integrates the algorithmic mechanisms of rollout scaling with the system-level throughput gains of NVFP4. This synergistic algorithm-hardware design effectively accelerates the rollout phase while reserving high-fidelity samples for optimization. We empirically demonstrate that our framework maintains the training integrity of BF16 precision pipeline while fully exploiting the throughput gains enabled by FP4 arithmetic. Extensive experiments across SANA, FLUX.1, and SD3.5-L substantiate that our approach delivers superior alignment performance across multiple metrics while accelerating training convergence by up to $4.64\times$, unlocking the power of massive rollout scaling at a fraction of the cost.

In this paper, we study a vehicle-to-infrastructure (V2I) system where distributed base stations (BSs) acting as road-side units (RSUs) collect multimodal (wireless and visual) data from moving vehicles. We consider a decentralized rate maximization problem, where each RSU relies on its local observations to optimize its resources, while all RSUs must collaborate to guarantee favorable network performance. We recast this problem as a distributed multi-agent reinforcement learning (MARL) problem, by incorporating rotation symmetries in terms of vehicles' locations. To exploit these symmetries, we propose a novel self-supervised learning framework where each BS agent aligns the latent features of its multimodal observation to extract the positions of the vehicles in its local region. Equipped with this sensing data at each RSU, we train an equivariant policy network using a graph neural network (GNN) with message passing layers, such that each agent computes its policy locally, while all agents coordinate their policies via a signaling scheme that overcomes partial observability and guarantees the equivariance of the global policy. We present numerical results carried out in a simulation environment, where ray-tracing and computer graphics are used to collect wireless and visual data. Results show the generalizability of our self-supervised and multimodal sensing approach, achieving more than two-fold accuracy gains over baselines, and the efficiency of our equivariant MARL training, attaining more than 50% performance gains over standard approaches.

MLLMs require high-resolution visual inputs for fine-grained tasks like document understanding and dense scene perception. However, current global resolution scaling paradigms indiscriminately flood the quadratic self-attention mechanism with visually redundant tokens, severely bottlenecking inference throughput while ignoring spatial sparsity and query intent. To overcome this, we propose Q-Zoom, a query-aware adaptive high-resolution perception framework that operates in an efficient coarse-to-fine manner. First, a lightweight Dynamic Gating Network safely bypasses high-resolution processing when coarse global features suffice. Second, for queries demanding fine-grained perception, a Self-Distilled Region Proposal Network (SD-RPN) precisely localizes the task-relevant Region-of-Interest (RoI) directly from intermediate feature spaces. To optimize these modules efficiently, the gating network uses a consistency-aware generation strategy to derive deterministic routing labels, while the SD-RPN employs a fully self-supervised distillation paradigm. A continuous spatio-temporal alignment scheme and targeted fine-tuning then seamlessly fuse the dense local RoI with the coarse global layout. Extensive experiments demonstrate that Q-Zoom establishes a dominant Pareto frontier. Using Qwen2.5-VL-7B as a primary testbed, Q-Zoom accelerates inference by 2.52 times on Document & OCR benchmarks and 4.39 times in High-Resolution scenarios while matching the baseline's peak accuracy. Furthermore, when configured for maximum perceptual fidelity, Q-Zoom surpasses the baseline's peak performance by 1.1% and 8.1% on these respective benchmarks. These robust improvements transfer seamlessly to Qwen3-VL, LLaVA, and emerging RL-based thinking-with-image models. Project page is available at https://yuhengsss.github.io/Q-Zoom/.

As Large Language Models reshape the global labor market, policymakers and workers need empirical data on which occupational skills may be most susceptible to automation. We present the Skill Automation Feasibility Index (SAFI), benchmarking four frontier LLMs -- LLaMA 3.3 70B, Mistral Large, Qwen 2.5 72B, and Gemini 2.5 Flash -- across 263 text-based tasks spanning all 35 skills in the U.S. Department of Labor's O*NET taxonomy (1,052 total model calls, 0% failure rate). Cross-referencing with real-world AI adoption data from the Anthropic Economic Index (756 occupations, 17,998 tasks), we propose an AI Impact Matrix -- an interpretive framework that positions skills along four quadrants: High Displacement Risk, Upskilling Required, AI-Augmented, and Lower Displacement Risk. Key findings: (1) Mathematics (SAFI: 73.2) and Programming (71.8) receive the highest automation feasibility scores; Active Listening (42.2) and Reading Comprehension (45.5) receive the lowest; (2) a "capability-demand inversion" where skills most demanded in AI-exposed jobs are those LLMs perform least well at in our benchmark; (3) 78.7% of observed AI interactions are augmentation, not automation; (4) all four models converge to similar skill profiles (3.6-point spread), suggesting that text-based automation feasibility may be more skill-dependent than model-dependent. SAFI measures LLM performance on text-based representations of skills, not full occupational execution. All data, code, and model responses are open-sourced.

Large language models (LLMs) have demonstrated remarkable capabilities across diverse domains, yet their adaptation to specialized fields remains challenging, particularly for non-English languages. This study investigates domain-adaptive pre-training (DAPT) as a strategy for specializing small to mid-sized LLMs in the French biomedical domain through continued pre-training. We address two key research questions: the viability of specialized continued pre-training for domain adaptation and the relationship between domain-specific performance gains and general capability degradation. Our contributions include the release of a fully open-licensed French biomedical corpus suitable for commercial and open-source applications, the training and release of specialized French biomedical LLMs, and novel insights for DAPT implementation. Our methodology encompasses the collection and refinement of high-quality French biomedical texts, the exploration of causal language modeling approaches using DAPT, and conducting extensive comparative evaluations. Our results cast doubt on the efficacy of DAPT, in contrast to previous works, but we highlight its viability in smaller-scale, resource-constrained scenarios under the right conditions. Findings in this paper further suggest that model merging post-DAPT is essential to mitigate generalization trade-offs, and in some cases even improves performance on specialized tasks at which the DAPT was directed.

A fundamental obstacle to causal discovery from text is the lack of causally annotated text data for use as ground truth, due to high annotation costs. This motivates an important task of generating text with causal graph annotations. Early template-based generation methods sacrifice text naturalness in exchange for high causal graph annotation accuracy. Recent Large Language Model (LLM)-dependent methods directly generate natural text from target graphs through LLMs, but do not guarantee causal graph annotation accuracy. Therefore, we propose iTAG, which performs real-world concept assignment to nodes before converting causal graphs into text in existing LLM-dependent methods. iTAG frames this process as an inverse problem with the causal graph as the target, iteratively examining and refining concept selection through Chain-of-Thought (CoT) reasoning so that the induced relations between concepts are as consistent as possible with the target causal relationships described by the causal graph. iTAG demonstrates both extremely high annotation accuracy and naturalness across extensive tests, and the results of testing text-based causal discovery algorithms with the generated data show high statistical correlation with real-world data. This suggests that iTAG-generated data can serve as a practical surrogate for scalable benchmarking of text-based causal discovery algorithms.

Conventional career guidance platforms rely on static, text-driven interfaces that struggle to engage users or deliver personalised, evidence-based insights. Although Computer-Assisted Career Guidance Systems have evolved since the 1960s, they remain limited in interactivity and pay little attention to the narrative dimensions of career development. We introduce XR-CareerAssist, a platform that unifies Extended Reality (XR) with several Artificial Intelligence (AI) modules to deliver immersive, multilingual career guidance. The system integrates Automatic Speech Recognition for voice-driven interaction, Neural Machine Translation across English, Greek, French, and Italian, a Langchain-based conversational Training Assistant for personalised dialogue, a BLIP-based Vision-Language model for career visualisations, and AWS Polly Text-to-Speech delivered through an interactive 3D avatar. Career trajectories are rendered as dynamic Sankey diagrams derived from a repository of more than 100,000 anonymised professional profiles. The application was built in Unity for Meta Quest 3, with backend services hosted on AWS. A pilot evaluation at the University of Exeter with 23 participants returned 95.6% speech recognition accuracy, 78.3% overall user satisfaction, and 91.3% favourable ratings for system responsiveness, with feedback informing subsequent improvements to motion comfort, audio clarity, and text legibility. XR-CareerAssist demonstrates how the fusion of XR and AI can produce more engaging, accessible, and effective career development tools, with the integration of five AI modules within a single immersive environment yielding a multimodal interaction experience that distinguishes it from existing career guidance platforms.

The field of cybersecurity is confronted with two interrelated challenges: a worldwide deficit of qualified practitioners and ongoing human-factor weaknesses that account for the bulk of security incidents. To tackle these issues, we present SentinelSphere, a platform driven by artificial intelligence that unifies machine learning-based threat identification with security training powered by a Large Language Model (LLM). The detection module uses an Enhanced Deep Neural Network (DNN) trained on the CIC-IDS2017 and CIC-DDoS2019 benchmark datasets, enriched with novel HTTP-layer feature engineering that captures application level attack signatures. For the educational component, we deploy a quantised variant of Phi-4 model (Q4_K_M), fine-tuned for the cybersecurity domain, enabling deployment on commodity hardware requiring only 16 GB of RAM without dedicated GPU resources. Experimental results show that the Enhanced DNN attains high detection accuracy while substantially lowering false positives relative to baseline models, and maintains strong recall across critical attack categories such as DDoS, brute force, and web-based exploits. Validation workshops involving industry professionals and university students confirmed that the Traffic Light visualisation system and conversational AI assistant are both intuitive and effective for users without technical backgrounds. SentinelSphere illustrates that coupling intelligent threat detection with adaptive, LLM-driven security education can meaningfully address both technical and human-factor cybersecurity vulnerabilities within a single, cohesive framework.

Data leakage is the inadvertent transfer of information between training and evaluation datasets that poses a subtle, yet critical, risk to the reliability of machine learning (ML) models in safety-critical systems such as automotive perception. While leakage is widely recognized in research, little is known about how industrial practitioners actually perceive and manage it in practice. This study investigates practitioners' knowledge, experiences, and mitigation strategies around data leakage through ten semi-structured interviews with system design, development, and verification engineers working on automotive perception functions development. Using reflexive thematic analysis, we identify that knowledge of data leakage is widespread and fragmented along role boundaries: ML engineers conceptualize it as a data-splitting or validation issue, whereas design and verification roles interpret it in terms of representativeness and scenario coverage. Detection commonly arises through generic considerations and observed performance anomalies rather than implying specific tools. However, data leakage prevention is more commonly practiced, which depends mostly on experience and knowledge sharing. These findings suggest that leakage control is a socio-technical coordination problem distributed across roles and workflows. We discuss implications for ML reliability engineering, highlighting the need for shared definitions, traceable data practices, and continuous cross-role communication to institutionalize data leakage awareness within automotive ML development.

Epithelial tissues dynamically reshape through local mechanical interactions among cells, a process well captured by vertex models. Yet their many tunable parameters make inference and optimization challenging, motivating computational frameworks that flexibly model and learn tissue mechanics. We introduce VertAX, a differentiable JAX-based framework for vertex-modeling of confluent epithelia. VertAX provides automatic differentiation, GPU acceleration, and end-to-end bilevel optimization for forward simulation, parameter inference, and inverse mechanical design. Users can define arbitrary energy and cost functions in pure Python, enabling seamless integration with machine-learning pipelines. We demonstrate VertAX on three representative tasks: (i) forward modeling of tissue morphogenesis, (ii) mechanical parameter inference, and (iii) inverse design of tissue-scale behaviors. We benchmark three differentiation strategies-automatic differentiation, implicit differentiation, and equilibrium propagation-showing that the latter can approximate gradients using repeated forward, adjoint-free simulations alone, offering a simple route for extending inverse biophysical problems to non-differentiable simulators with limited additional engineering effort.

Deep convolutional neural networks achieve remarkable performance by exhaustively processing dense spatial feature maps, yet this brute-force strategy introduces significant computational redundancy and encourages reliance on spurious background correlations. As a result, modern vision models remain brittle and difficult to interpret. We propose Energy-Regularized Spatial Masking (ERSM), a novel framework that reformulates feature selection as a differentiable energy minimization problem. By embedding a lightweight Energy-Mask Layer inside standard convolutional backbones, each visual token is assigned a scalar energy composed of two competing forces: an intrinsic Unary importance cost and a Pairwise spatial coherence penalty. Unlike prior pruning methods that enforce rigid sparsity budgets or rely on heuristic importance scores, ERSM allows the network to autonomously discover an optimal information-density equilibrium tailored to each input. We validate ERSM on convolutional architectures and demonstrate that it produces emergent sparsity, improved robustness to structured occlusion, and highly interpretable spatial masks, while preserving classification accuracy. Furthermore, we show that the learned energy ranking significantly outperforms magnitude-based pruning in deletion-based robustness tests, revealing ERSM as an intrinsic denoising mechanism that isolates semantic object regions without pixel-level supervision.

Neural operators have emerged as powerful surrogates for dynamical systems due to their grid-invariant properties and computational efficiency. However, the Fourier-based neural operator framework inherently truncates high-frequency components in spectral space, resulting in the loss of small-scale structures and degraded prediction quality at high resolutions when trained on low-resolution data. While diffusion-based enhancement methods can recover multi-scale features, they introduce substantial inference overhead that undermines the efficiency advantage of neural operators. In this work, we introduce \textbf{M}eanFlow-\textbf{E}nhanced \textbf{N}eural \textbf{O}perators (MENO), a novel framework that achieves accurate all-scale predictions with minimal inference cost. By leveraging the improved MeanFlow method, MENO restores both small-scale details and large-scale dynamics with superior physical fidelity and statistical accuracy. We evaluate MENO on three challenging dynamical systems, including phase-field dynamics, 2D Kolmogorov flow, and active matter dynamics, at resolutions up to 256$\times$256. Across all benchmarks, MENO improves the power spectrum density accuracy by up to a factor of 2 compared to baseline neural operators while achieving 12$\times$ faster inference than the state-of-the-art Diffusion Denoising Implicit Model (DDIM)-enhanced counterparts, effectively bridging the gap between accuracy and efficiency. The flexibility and efficiency of MENO position it as an efficient surrogate model for scientific machine learning applications where both statistical integrity and computational efficiency are paramount.

In Milner's seminal book on communication and concurrency introducing CCS, a process algebra inherently non-deterministic, chapter 11 was completely devoted to introduce the notion of determinacy and confluence in order to identify a subcalculus of CCS in which all definable agents are confluent. At the same time, or shortly later, determinate semantics were given for programming languages that reconcile concurrency and determinacy, such as Esterel by Berry and Gonthier, or SL by Boussinot and de Simone. These dedicated semantics do not easily map to Milner's confluence theory for CCS, which is unable to express causality and shared memory multi-threading with reaction to absence in a compositional way. We present an extension of CCS with priority-guarded actions and clocks, and we exploit the added expressiveness to enrich Milner's original notion of confluence by the new concept of coherence which permits us to encode, in a compositional fashion, synchronous programming languages such as Esterel.

Large Speech Language Models (LSLMs) typically operate at high token rates (tokens/s) to ensure acoustic fidelity, yet this results in sequence lengths that far exceed the underlying semantic content, incurring prohibitive inference costs. In this paper, we empirically revisit the necessity of such granular token-level processing. Through layer-wise oracle interventions, we unveil a structured redundancy hierarchy: while shallow layers encode essential acoustic details, deep layers exhibit extreme redundancy, allowing for aggressive compression. Motivated by these findings, we introduce Affinity Pooling, a training-free, similarity-based token merging mechanism. By strategically applying this method at both input and deep layers, we effectively compress speech representations without compromising semantic information. Extensive evaluations across three tasks demonstrate that our approach reduces prefilling FLOPs by 27.48\% while maintaining competitive accuracy. Practical deployment further confirms significant efficiency gains, yielding up to $\sim$1.7$\times$ memory savings and $\sim$1.1$\times$ faster time-to-first-token on long utterances. Our results challenge the necessity of fully distinct token representations, providing new perspectives on LSLM efficiency.

Physical adversarial attacks are increasingly studied in settings that resemble deployed surveillance systems rather than isolated image benchmarks. In these settings, person detection, multi-object tracking, visible--infrared sensing, and the practical form of the attack carrier all matter at once. This changes how the literature should be read. A perturbation that suppresses a detector in one frame may have limited practical effect if identity is recovered over time; an RGB-only result may say little about night-time systems that rely on visible and thermal inputs together; and a conspicuous patch can imply a different threat model from a wearable or selectively activated carrier. This paper reviews physical attacks from that surveillance-oriented viewpoint. Rather than attempting a complete catalogue of all physical attacks in computer vision, we focus on the technical questions that become central in surveillance: temporal persistence, sensing modality, carrier realism, and system-level objective. We organize prior work through a four-part taxonomy and discuss how recent results on multi-object tracking, dual-modal visible--infrared evasion, and controllable clothing reflect a broader change in the field. We also summarize evaluation practices and unresolved gaps, including distance robustness, camera-pipeline variation, identity-level metrics, and activation-aware testing. The resulting picture is that surveillance robustness cannot be judged reliably from isolated per-frame benchmarks alone; it has to be examined as a system problem unfolding over time, across sensors, and under realistic physical deployment constraints.

Clustering in high-dimensional settings with severe feature noise remains challenging, especially when only a small subset of dimensions is informative and the final number of clusters is not specified in advance. In such regimes, partition recovery, feature relevance learning, and structural adaptation are tightly coupled, and standard likelihood-based methods can become unstable or overly sensitive to noisy dimensions. We propose DIVI, a data-informed variational clustering framework that combines global feature gating with split-based adaptive structure growth. DIVI uses informative prior initialization to stabilize optimization, learns feature relevance in a differentiable manner, and expands model complexity only when local diagnostics indicate underfit. Beyond clustering performance, we also examine runtime scalability and parameter sensitivity in order to clarify the computational and practical behavior of the framework. Empirically, we find that DIVI performs competitively under severe feature noise, remains computationally feasible, and yields interpretable feature-gating behavior, while also exhibiting conservative growth and identifiable failure regimes in challenging settings. Overall, DIVI is best viewed as a practical variational clustering framework for noisy high-dimensional data rather than as a fully Bayesian generative solution.

Skin-toned emojis are crucial for fostering personal identity and social inclusion in online communication. As AI models, particularly Large Language Models (LLMs), increasingly mediate interactions on web platforms, the risk that these systems perpetuate societal biases through their representation of such symbols is a significant concern. This paper presents the first large-scale comparative study of bias in skin-toned emoji representations across two distinct model classes. We systematically evaluate dedicated emoji embedding models (emoji2vec, emoji-sw2v) against four modern LLMs (Llama, Gemma, Qwen, and Mistral). Our analysis first reveals a critical performance gap: while LLMs demonstrate robust support for skin tone modifiers, widely-used specialized emoji models exhibit severe deficiencies. More importantly, a multi-faceted investigation into semantic consistency, representational similarity, sentiment polarity, and core biases uncovers systemic disparities. We find evidence of skewed sentiment and inconsistent meanings associated with emojis across different skin tones, highlighting latent biases within these foundational models. Our findings underscore the urgent need for developers and platforms to audit and mitigate these representational harms, ensuring that AI's role on the web promotes genuine equity rather than reinforcing societal biases.

Issue resolution aims to automatically generate patches from given issue descriptions and has attracted significant attention with the rapid advancement of large language models (LLMs). However, due to the complexity of software issues and codebases, LLM-generated patches often fail to resolve corresponding issues. Although various advanced techniques have been proposed with carefully designed tools and workflows, they typically treat issue descriptions as direct inputs and largely overlook their quality (e.g., missing critical context or containing ambiguous information), which hinders LLMs from accurate understanding and resolution. To address this limitation, we draw on principles from software requirements engineering and propose REAgent, a requirement-driven LLM agent framework that introduces issue-oriented requirements as structured task specifications to better guide patch generation. Specifically, REAgent automatically constructs structured and information-rich issue-oriented requirements, identifies low-quality requirements, and iteratively refines them to improve patch correctness. We conduct comprehensive experiments on three widely used benchmarks using two advanced LLMs, comparing against five representative or state-of-the-art baselines. The results demonstrate that REAgent consistently outperforms all baselines, achieving an average improvement of 17.40% in terms of the number of successfully-resolved issues (% Resolved).

BACKGROUND: Recent studies have shown that domain-adapted large language models (LLMs) do not consistently outperform general-purpose counterparts on standard medical benchmarks, raising questions about the need for specialized clinical adaptation. METHODS: We systematically compare general and clinical LLMs on a diverse set of multiple choice clinical question answering tasks in English and Spanish. We introduce a perturbation based evaluation benchmark that probes model robustness, instruction following, and sensitivity to adversarial variations. Our evaluation includes, one-step and two-step question transformations, multi prompt testing and instruction guided assessment. We analyze a range of state-of-the-art clinical models and their general-purpose counterparts, focusing on Llama 3.1-based models. Additionally, we introduce Marmoka, a family of lightweight 8B-parameter clinical LLMs for English and Spanish, developed via continual domain-adaptive pretraining on medical corpora and instructions. RESULTS: The experiments show that clinical LLMs do not consistently outperform their general purpose counterparts on English clinical tasks, even under the proposed perturbation based benchmark. However, for the Spanish subsets the proposed Marmoka models obtain better results compared to Llama. CONCLUSIONS: Our results show that, under current short-form MCQA benchmarks, clinical LLMs offer only marginal and unstable improvements over general-purpose models in English, suggesting that existing evaluation frameworks may be insufficient to capture genuine medical expertise. We further find that both general and clinical models exhibit substantial limitations in instruction following and strict output formatting. Finally, we demonstrate that robust medical LLMs can be successfully developed for low-resource languages such as Spanish, as evidenced by the Marmoka models.

Interactive medical dialogue benchmarks have shown that LLM diagnostic accuracy degrades significantly when interacting with non-cooperative patients, yet existing approaches either apply adversarial behaviors without graded severity or case-specific grounding, or reduce patient non-cooperation to a single ungraded axis, and none analyze cross-dimension interactions. We introduce MedDialBench, a benchmark enabling controlled, dose-response characterization of how individual patient behavior dimensions affect LLM diagnostic robustness. It decomposes patient behavior into five dimensions -- Logic Consistency, Health Cognition, Expression Style, Disclosure, and Attitude -- each with graded severity levels and case-specific behavioral scripts. This controlled factorial design enables graded sensitivity analysis, dose-response profiling, and cross-dimension interaction detection. Evaluating five frontier LLMs across 7,225 dialogues (85 cases x 17 configurations x 5 models), we find a fundamental asymmetry: information pollution (fabricating symptoms) produces 1.7-3.4x larger accuracy drops than information deficit (withholding information), and fabricating is the only configuration achieving statistical significance across all five models (McNemar p < 0.05). Among six dimension combinations, fabricating is the sole driver of super-additive interaction: all three fabricating-involving pairs produce O/E ratios of 0.70-0.81 (35-44% of eligible cases fail under the combination despite succeeding under each dimension alone), while all non-fabricating pairs show purely additive effects (O/E ~ 1.0). Inquiry strategy moderates deficit but not pollution: exhaustive questioning recovers withheld information, but cannot compensate for fabricated inputs. Models exhibit distinct vulnerability profiles, with worst-case drops ranging from 38.8 to 54.1 percentage points.

Long-term memory is critical for dialogue systems that support continuous, sustainable, and personalized interactions. However, existing methods rely on continuous summarization or OpenIE-based graph construction paired with fixed Top-\textit{k} retrieval, leading to limited adaptability across query categories and high computational overhead. In this paper, we propose HingeMem, a boundary-guided long-term memory that operationalizes event segmentation theory to build an interpretable indexing interface via boundary-triggered hyperedges over four elements: person, time, location, and topic. When any such element changes, HingeMem draws a boundary and writes the current segment, thereby reducing redundant operations and preserving salient context. To enable robust and efficient retrieval under diverse information needs, HingeMem introduces query-adaptive retrieval mechanisms that jointly decide (a) \textit{what to retrieve}: determine the query-conditioned routing over the element-indexed memory; (b) \textit{how much to retrieve}: control the retrieval depth based on the estimated query type. Extensive experiments across LLM scales (from 0.6B to production-tier models; \textit{e.g.}, Qwen3-0.6B to Qwen-Flash) on LOCOMO show that HingeMem achieves approximately $20\%$ relative improvement over strong baselines without query categories specification, while reducing computational cost (68\%$\downarrow$ question answering token cost compared to HippoRAG2). Beyond advancing memory modeling, HingeMem's adaptive retrieval makes it a strong fit for web applications requiring efficient and trustworthy memory over extended interactions.

In this paper, we propose using Learning from Answer Sets to approximate black-box models, such as Neural Networks (NN), in the specific case of learning user preferences. We specifically explore the use of ILASP (Inductive Learning of Answer Set Programs) to approximate preference learning systems through weak constraints. We have created a dataset on user preferences over a set of recipes, which is used to train the NNs that we aim to approximate with ILASP. Our experiments investigate ILASP both as a global and a local approximator of the NNs. These experiments address the challenge of approximating NNs working on increasingly high-dimensional feature spaces while achieving appropriate fidelity on the target model and limiting the increase in computational time. To handle this challenge, we propose a preprocessing step that exploits Principal Component Analysis to reduce the dataset's dimensionality while keeping our explanations transparent. Under consideration for publication in Theory and Practice of Logic Programming (TPLP).

The problem of solving Markov decision processes under function approximation remains a fundamental challenge, even under linear function approximation settings. A key difficulty arises from a geometric mismatch: while the Bellman optimality operator is contractive in the Linfty-norm, commonly used objectives such as projected value iteration and Bellman residual minimization rely on L2-based formulations. To enable gradient-based optimization, we consider a soft formulation of Bellman residual minimization and extend it to a generalized weighted Lp -norm. We show that this formulation aligns the optimization objective with the contraction geometry of the Bellman operator as p increases, and derive corresponding performance error bounds. Our analysis provides a principled connection between residual minimization and Bellman contraction, leading to improved control of error propagation while remaining compatible with gradient-based optimization.

Quantization is an effective way to reduce the memory cost of large-scale model training. However, most existing methods adopt fixed-precision policies, which ignore the fact that optimizer-state distributions vary significantly across layers and training steps. Such uniform designs often introduce noticeable accuracy degradation. To move beyond fixed quantization, we propose STQuant, a distributed training framework that reduces the memory footprint of optimizer states via dynamic precision allocation across layers, state variables, and training steps, while maintaining model quality. Naively applying dynamic quantization during training is challenging for two reasons. First, optimizer states are numerically sensitive, and quantization noise can destabilize quality. Second, jointly considering multiple states and layers induces a large combinatorial search space. STQuant addresses these challenges with two key techniques: 1) a provably near-optimal factor selection strategy that accurately identifies the most influential factors for precision adaptation. 2) a dynamic transition decision algorithm that reduces the search cost from exponential to linear complexity. Experiments on GPT-2 and ViT show that STQuant reduces optimizer-state memory by 84.4%, achieving an average bit-width of as low as 5.1 bits, compared with existing solutions. Moreover, STQuant incurs only O(N/K) computational overhead and requires O(1) extra space.

Large reasoning models have recently demonstrated strong performance on complex tasks that require long chain-of-thought reasoning, through supervised fine-tuning on large-scale and high-quality datasets. To construct such datasets, existing pipelines generate long reasoning data from more capable Large Language Models (LLMs) and apply manually heuristic or naturalness-based selection methods to filter high-quality samples. Despite the proven effectiveness of naturalness-based data selection, which ranks data by the average log probability assigned by LLMs, our analysis shows that, when applied to LLM reasoning datasets, it systematically prefers samples with longer reasoning steps (i.e., more tokens per step) rather than higher-quality ones, a phenomenon we term step length confounding. Through quantitative analysis, we attribute this phenomenon to low-probability first tokens in reasoning steps; longer steps dilute their influence, thereby inflating the average log probabilities. To address this issue, we propose two variant methods: ASLEC-DROP, which drops first-token probabilities when computing average log probability, and ASLEC-CASL, which applies a causal debiasing regression to remove the first tokens' confounding effect. Experiments across four LLMs and five evaluation benchmarks demonstrate the effectiveness of our approach in mitigating the step length confounding problem.

As high quality public data becomes scarce, Federated Learning (FL) provides a vital pathway to leverage valuable private user data while preserving privacy. However, real-world client data often contains toxic or unsafe information. This leads to a critical issue we define as unintended data poisoning, which can severely damage the safety alignment of global models during federated alignment. To address this, we propose FedDetox, a robust framework tailored for Small Language Models (SLMs) on resource-constrained edge devices. We first employ knowledge distillation to transfer sophisticated safety alignment capabilities from large scale safety aligned teacher models into light weight student classifiers suitable for resource constrained edge devices. Specifically, during federated learning for human preference alignment, the edge client identifies unsafe samples at the source and replaces them with refusal templates, effectively transforming potential poisons into positive safety signals. Experiments demonstrate that our approach preserves model safety at a level comparable to centralized baselines without compromising general utility.

Vision-language models (VLMs) predominantly rely on autoregressive decoding, which generates tokens one at a time and fundamentally limits inference throughput. This limitation is especially acute in physical AI scenarios such as robotics and autonomous driving, where VLMs are deployed on edge devices at batch size one, making AR decoding memory-bandwidth-bound and leaving hardware parallelism underutilized. While block-wise discrete diffusion has shown promise for parallel text generation, extending it to VLMs remains challenging due to the need to jointly handle continuous visual representations and discrete text tokens while preserving pretrained multimodal capabilities. We present Fast-dVLM, a block-diffusion-based VLM that enables KV-cache-compatible parallel decoding and speculative block decoding for inference acceleration. We systematically compare two AR-to-diffusion conversion strategies: a two-stage approach that first adapts the LLM backbone with text-only diffusion fine-tuning before multimodal training, and a direct approach that converts the full AR VLM in one stage. Under comparable training budgets, direct conversion proves substantially more efficient by leveraging the already multimodally aligned VLM; we therefore adopt it as our recommended recipe. We introduce a suite of multimodal diffusion adaptations, block size annealing, causal context attention, auto-truncation masking, and vision efficient concatenation, that collectively enable effective block diffusion in the VLM setting. Extensive experiments across 11 multimodal benchmarks show Fast-dVLM matches its autoregressive counterpart in generation quality. With SGLang integration and FP8 quantization, Fast-dVLM achieves over 6x end-to-end inference speedup over the AR baseline.

Current LLM-based services typically require users to submit raw text regardless of its sensitivity. While intuitive, such practice introduces substantial privacy risks, as unauthorized access may expose personal, medical, or legal information. Although prior defenses strived to mitigate these risks, they often incur substantial computational overhead and degrade model performance. To overcome this privacy-efficiency trade-off, we introduce Privacy-Preserving Fine-Tuning (PPFT), a novel training pipeline that eliminates the need for transmitting raw prompt text while maintaining a favorable balance between privacy preservation and model utility for both clients and service providers. Our approach operates in two stages: first, we train a client-side encoder together with a server-side projection module and LLM, enabling the server to condition on k-pooled prompt embeddings instead of raw text; second, we fine-tune the projection module and LLM on private, domain-specific data using noise-injected embeddings, allowing effective adaptation without exposing plain text prompts and requiring access to the decoder's internal parameters. Extensive experiments on domain-specific and general benchmarks demonstrate that PPFT achieves a striking balance between privacy and utility, maintaining competitive performance with minimal degradation compared to noise-free upper bounds.

Synthetic data rephrasing has emerged as a powerful technique for enhancing knowledge acquisition during large language model (LLM) pretraining. However, existing approaches operate at the single-document level, rewriting individual web pages in isolation. This confines synthesized examples to intra-document knowledge, missing cross-document relationships and leaving facts with limited associative context. We propose WRAP++ (Web discoveRy Amplified Pretraining), which amplifies the associative context of factual knowledge by discovering cross-document relationships from web hyperlinks and synthesizing joint QA over each discovered document pair. Concretely, WRAP++ discovers high-confidence relational motifs including dual-links and co-mentions, and synthesizes QA that requires reasoning across both documents. This produces relational knowledge absent from either source document alone, creating diverse entry points to the same facts. Because the number of valid entity pairs grows combinatorially, this discovery-driven synthesis also amplifies data scale far beyond single-document rewriting. Instantiating WRAP++ on Wikipedia, we amplify ~8.4B tokens of raw text into 80B tokens of cross-document QA data. On SimpleQA, OLMo-based models at both 7B and 32B scales trained with WRAP++ substantially outperform single-document approaches and exhibit sustained scaling gains, underscoring the advantage of cross-document knowledge discovery and amplification.

Environmental, Social, and Governance (ESG) considerations are increasingly integral to assessing corporate performance, reputation, and long-term sustainability. Yet, reliable ESG ratings remain limited for smaller companies and emerging markets. We introduce the first publicly available Slovene ESG sentiment dataset and a suite of models for automatic ESG sentiment detection. The dataset, derived from the MaCoCu Slovene news collection, combines large language model (LLM)-assisted filtering with human annotation of company-related ESG content. We evaluate the performance of monolingual (SloBERTa) and multilingual (XLM-R) models, embedding-based classifiers (TabPFN), hierarchical ensemble architectures, and large language models. Results show that LLMs achieve the strongest performance on Environmental (Gemma3-27B, F1-macro: 0.61) and Social aspects (gpt-oss 20B, F1-macro: 0.45), while fine-tuned SloBERTa is the best model on Governance classification (F1-macro: 0.54). We then show in a small case study how the best-preforming classifier (gpt-oss) can be applied to investigate ESG aspects for selected companies across a long time frame.

Large language models (LLMs) can generate persuasive narratives at scale, raising concerns about their potential use in disinformation campaigns. Assessing this risk ultimately requires understanding how readers receive such content. In practice, however, LLM judges are increasingly used as a low-cost substitute for direct human evaluation, even though whether they faithfully track reader responses remains unclear. We recast evaluation in this setting as a proxy-validity problem and audit LLM judges against human reader responses. Using 290 aligned articles, 2,043 paired human ratings, and outputs from eight frontier judges, we examine judge--human alignment in terms of overall scoring, item-level ordering, and signal dependence. We find persistent judge--human gaps throughout. Relative to humans, judges are typically harsher, recover item-level human rankings only weakly, and rely on different textual signals, placing more weight on logical rigour while penalizing emotional intensity more strongly. At the same time, judges agree far more with one another than with human readers. These results suggest that LLM judges form a coherent evaluative group that is much more aligned internally than it is with human readers, indicating that internal agreement is not evidence of validity as a proxy for reader response.

We present SemEval-2026 Task 9, a shared task on online polarization detection, covering 22 languages and comprising over 110K annotated instances. Each data instance is multi-labeled with the presence of polarization, polarization type, and polarization manifestation. Participants were asked to predict labels in three sub-tasks: (1) detecting the presence of polarization, (2) identifying the type of polarization, and (3) recognizing the polarization manifestation. The three tasks attracted over 1,000 participants worldwide and more than 10k submission on Codabench. We received final submissions from 67 teams and 73 system description papers. We report the baseline results and analyze the performance of the best-performing systems, highlighting the most common approaches and the most effective methods across different subtasks and languages. The dataset of this task is publicly available.

While traditional tree-based ensemble methods have long dominated tabular tasks, deep neural networks and emerging foundation models have challenged this primacy, yet no consensus exists on a universally superior paradigm. Existing benchmarks typically contain fewer than 100 datasets, raising concerns about evaluation sufficiency and potential selection biases. To address these limitations, we introduce OmniTabBench, the largest tabular benchmark to date, comprising 3030 datasets spanning diverse tasks that are comprehensively collected from diverse sources and categorized by industry using large language models. We conduct an unprecedented large-scale empirical evaluation of state-of-the-art models from all model families on OmniTabBench, confirming the absence of a dominant winner. Furthermore, through a decoupled metafeature analysis, which examines individual properties such as dataset size, feature types, feature and target skewness/kurtosis, we elucidate conditions favoring specific model categories, providing clearer, more actionable guidance than prior compound-metric studies.

Large Language Models (LLMs) have demonstrated impressive capabilities in long-form generation, yet their application is hindered by the hallucination problem. While Uncertainty Quantification (UQ) is essential for assessing reliability, the complex structure makes reliable aggregation across heterogeneous themes difficult, in addition, existing methods often overlook the nuance of neutral information and suffer from the high computational cost of fine-grained decomposition. To address these challenges, we propose AGSC (Adaptive Granularity and GMM-based Semantic Clustering), a UQ framework tailored for long-form generation. AGSC first uses NLI neutral probabilities as triggers to distinguish irrelevance from uncertainty, reducing unnecessary computation. It then applies Gaussian Mixture Model (GMM) soft clustering to model latent semantic themes and assign topic-aware weights for downstream aggregation. Experiments on BIO and LongFact show that AGSC achieves state-of-the-art correlation with factuality while reducing inference time by about 60% compared to full atomic decomposition.

Skill-based agent systems tackle complex tasks by composing reusable skills, improving modularity and scalability while introducing a largely unexamined security attack surface. We propose SkillTrojan, a backdoor attack that targets skill implementations rather than model parameters or training data. SkillTrojan embeds malicious logic inside otherwise plausible skills and leverages standard skill composition to reconstruct and execute an attacker-specified payload. The attack partitions an encrypted payload across multiple benign-looking skill invocations and activates only under a predefined trigger. SkillTrojan also supports automated synthesis of backdoored skills from arbitrary skill templates, enabling scalable propagation across skill-based agent ecosystems. To enable systematic evaluation, we release a dataset of 3,000+ curated backdoored skills spanning diverse skill patterns and trigger-payload configurations. We instantiate SkillTrojan in a representative code-based agent setting and evaluate both clean-task utility and attack success rate. Our results show that skill-level backdoors can be highly effective with minimal degradation of benign behavior, exposing a critical blind spot in current skill-based agent architectures and motivating defenses that explicitly reason about skill composition and execution. Concretely, on EHR SQL, SkillTrojan attains up to 97.2% ASR while maintaining 89.3% clean ACC on GPT-5.2-1211-Global.

This paper presents CBM-Dual, the first silicon-proven digital chaotic dynamics processor (CDP) supporting both simulated annealing (SA) and reservoir computing (RC). CBM-Dual enables real-time decision-making and lightweight adaptation for autonomous Edge AI, employing the largest-scale fully connected 1024-neuron chaotic Boltzmann machine (CBM). To address the high computational and area costs of digital CDPs, we propose: 1) a CBM-specific scheduler that exploits an inherently low neuron flip rate to reduce multiply-accumulate operations by 99%, and 2) an efficient multiply splitting scheme that reduces the area by 59%. Fabricated in 65nm (12mm$^2$), CBM-Dual achieves simultaneous heterogeneous task execution and state-of-the-art energy efficiency, delivering $\times$25-54 and $\times$4.5 improvements in the SA and RC fields, respectively.

Multi-step Chain-of-Thought (CoT) has significantly advanced the mathematical reasoning capabilities of LLMs by leveraging explicit reasoning steps. However, the widespread adoption of Long CoT often results in sequence lengths that exceed manageable computational limits. While existing approaches attempt to alleviate this by reducing KV Cache redundancy via Markov chain-like structures, they introduce two critical limitations: inherent memorylessness (loss of context) and limited backward reasoning capability. To address these limitations, we propose a novel Chain-of-Thought framework based on Reversible Hierarchical Markov Chain, termed Cognitive Loop of Thought (CLoT), and a backward reasoning dataset CLoT-Instruct. In CLoT, problems are decomposed into sub-problems with hierarchical dependencies. Inspired by human cognitive processes, we introduce a backward verification mechanism at each hierarchical layer. Furthermore, we implement a pruning strategy: once higher-level sub-problems are verified, redundant lower-level sub-problems are pruned to maximize efficiency. This approach effectively mitigates error propagation and enhances reasoning robustness. Experiments on four mathematical benchmarks demonstrate the effectiveness of our method. Notably, on the AddSub dataset using GPT-4o-mini, CLoT achieves 99.0% accuracy, outperforming traditional CoT and CoT-SC by 4.1% and 2.9%, respectively.

Recent AI systems have achieved gold-medal-level performance on the International Mathematical Olympiad, demonstrating remarkable proficiency at competition-style problem solving. However, competition mathematics represents only a narrow slice of mathematical reasoning: problems are drawn from limited domains, require minimal advanced machinery, and can often reward insightful tricks over deep theoretical knowledge. We introduce \bench{}, a private benchmark of 25 expert-curated problems designed to evaluate AI systems on research-level mathematics that goes far beyond the olympiad frontier. Problems are authored by Ivy League mathematics professors, graduate students, and PhD-holding IMO medalists, and routinely took their authors weeks to solve independently. Each problem undergoes double-blind verification by two independent domain experts who must solve the problem from scratch, and yields a unique, closed-form solution assessed by programmatic verifiers. We evaluate frontier models as unconstrained research agents, with full access to coding tools, search, and open-ended reasoning, using an unbiased statistical estimator computed over 100 independent runs per problem. Our results reveal that all frontier models currently score below 10\%, exposing a substantial gap between olympiad-level problem solving and genuine research-level mathematical reasoning. By keeping the benchmark fully private, we ensure that measured performance reflects authentic mathematical capability rather than memorization of training data.

Algebraic reasoning remains one of the most informative stress tests for large language models, yet current benchmarks provide no mechanism for attributing failure to a specific cause. When a model fails an algebraic problem, a single accuracy score cannot reveal whether the expression was too deeply nested, the operator too uncommon, the intermediate state count too high, or the dependency chain too long. Prior work has studied individual failure modes in isolation, but no framework has varied each complexity factor independently under strict experimental control. No prior system has offered automatic generation and verification of problems of increasing complexity to track model progress over time. We introduce a nine-dimension algebraic complexity framework in which each factor is varied independently while all others are held fixed, with problem generation and verification handled by a parametric pipeline requiring no human annotation. Each dimension is grounded in a documented LLM failure mode and captures a structurally distinct aspect of algebraic difficulty, including expression nesting depth, simultaneous intermediate result count, sub-expression complexity, operator hardness, and dependent reasoning chain length. We evaluated seven instruction-tuned models spanning 8B to 235B parameters across all nine dimensions and find that working memory is the dominant scale-invariant bottleneck. Every model collapses between 20 and 30 parallel branches regardless of parameter count, pointing to a hard architectural constraint rather than a solvable capacity limitation. Our analysis further identifies a minimal yet diagnostically sufficient subset of five dimensions that together span the full space of documented algebraic failure modes, providing a complete complexity profile of a model's algebraic reasoning capacity.

Mixture-of-Experts (MoE) based large language models (LLMs) offer strong performance but suffer from high memory and computation costs. Weight binarization provides extreme efficiency, yet existing binary methods designed for dense LLMs struggle with MoE-specific issues, including cross-expert redundancy, task-agnostic importance estimation, and quantization-induced routing shifts. To this end, we propose MoBiE, the first binarization framework tailored for MoE-based LLMs. MoBiE is built on three core innovations: 1. using joint SVD decomposition to reduce cross-expert redundancy; 2. integrating global loss gradients into local Hessian metrics to enhance weight importance estimation; 3. introducing an error constraint guided by the input null space to mitigate routing distortion. Notably, MoBiE achieves these optimizations while incurring no additional storage overhead, striking a balance between efficiency and model performance. Extensive experiments demonstrate that MoBiE consistently outperforms state-of-the-art binary methods across multiple MoE-based LLMs and benchmarks. For example, on Qwen3-30B-A3B, MoBiE reduces perplexity by 52.2$\%$, improves average zero-shot performance by 43.4$\%$, achieves over 2 $\times$ inference speedup, and further shortens quantization time. The code is available at https://github.com/Kishon-zzx/MoBiE.

Latent variable models, including variational autoencoders (VAE), remain a central tool in modern deep generative modeling due to their scalability and a well-founded probabilistic formulation. These models rely on amortized variational inference to enable efficient posterior approximation, but this efficiency comes at the cost of a shared parametrization, giving rise to the amortization gap. We propose the instance-adaptive variational autoencoder (IA-VAE), an amortized variational inference framework in which a hypernetwork generates input-dependent modulations of a shared encoder. This enables input-specific adaptation of the inference model while preserving the efficiency of a single forward pass. By leveraging instance-specific parameter modulations, the proposed approach can achieve performance comparable to standard encoders with substantially fewer parameters, indicating a more efficient use of model capacity. Experiments on synthetic data, where the true posterior is known, show that IA-VAE yields more accurate posterior approximations and reduces the amortization gap. Similarly, on standard image benchmarks, IA-VAE consistently improves held-out ELBO over baseline VAEs, with statistically significant gains across multiple runs. These results suggest that increasing the flexibility of the inference parametrization through instance-adaptive modulation is a key factor in mitigating amortization-induced suboptimality in deep generative models.

Federated Learning (FL) enables decentralized model training across multiple clients without exposing private data, making it ideal for privacy-sensitive applications. However, in real-world FL scenarios, clients often hold data from distinct domains, leading to severe domain shift and degraded global model performance. To address this, prototype learning has been emerged as a promising solution, which leverages class-wise feature representations. Yet, existing methods face two key limitations: (1) Existing prototype-based FL methods typically construct a $\textit{single global prototype}$ per class by aggregating local prototypes from all clients without preserving domain information. (2) Current feature-prototype alignment is $\textit{domain-agnostic}$, forcing clients to align with global prototypes regardless of domain origin. To address these challenges, we propose Federated Domain-Aware Prototypes (FedDAP) to construct domain-specific global prototypes by aggregating local client prototypes within the same domain using a similarity-weighted fusion mechanism. These global domain-specific prototypes are then used to guide local training by aligning local features with prototypes from the same domain, while encouraging separation from prototypes of different domains. This dual alignment enhances domain-specific learning at the local level and enables the global model to generalize across diverse domains. Finally, we conduct extensive experiments on three different datasets: DomainNet, Office-10, and PACS to demonstrate the effectiveness of our proposed framework to address the domain shift challenges. The code is available at https://github.com/quanghuy6997/FedDAP.

Chain-of-Thought reasoning can enhance large language models, but it requires manually designed prompts to guide the model. Recently proposed CoT-decoding enables the model to generate CoT-style reasoning paths without prompts, but it is only applicable to problems with fixed answer sets. To address this limitation, we propose a general decoding strategy GCoT-decoding that extends applicability to a broader range of question-answering tasks. GCoT-decoding employs a two-stage branching method combining Fibonacci sampling and heuristic error backtracking to generate candidate decoding paths. It then splits each path into a reasoning span and an answer span to accurately compute path confidence, and finally aggregates semantically similar paths to identify a consensus answer, replacing traditional majority voting. We conduct extensive experiments on six datasets covering both fixed and free QA tasks. Our method not only maintains strong performance on fixed QA but also achieves significant improvements on free QA, demonstrating its generality.

Software documentation is crucial for repository comprehension. While Large Language Models (LLMs) advance documentation generation from code snippets to entire repositories, existing benchmarks have two key limitations: (1) they lack a holistic, repository-level assessment, and (2) they rely on unreliable evaluation strategies, such as LLM-as-a-judge, which suffers from vague criteria and limited repository-level knowledge. To address these issues, we introduce SWD-Bench, a novel benchmark for evaluating repository-level software documentation. Inspired by documentation-driven development, our strategy evaluates documentation quality by assessing an LLM's ability to understand and implement functionalities using the documentation, rather than by directly scoring it. This is measured through function-driven Question Answering (QA) tasks. SWD-Bench comprises three interconnected QA tasks: (1) Functionality Detection, to determine if a functionality is described; (2) Functionality Localization, to evaluate the accuracy of locating related files; and (3) Functionality Completion, to measure the comprehensiveness of implementation details. We construct the benchmark, containing 4,170 entries, by mining high-quality Pull Requests and enriching them with repository-level context. Experiments reveal limitations in current documentation generation methods and show that source code provides complementary value. Notably, documentation from the best-performing method improves the issue-solving rate of SWE-Agent by 20.00%, which demonstrates the practical value of high-quality documentation in supporting documentation-driven development.

Video-guided Multimodal Translation (VMT) has advanced significantly in recent years. However, most existing methods rely on locally aligned video segments paired one-to-one with subtitles, limiting their ability to capture global narrative context across multiple segments in long videos. To overcome this limitation, we propose a globally video-guided multimodal translation framework that leverages a pretrained semantic encoder and vector database-based subtitle retrieval to construct a context set of video segments closely related to the target subtitle semantics. An attention mechanism is employed to focus on highly relevant visual content, while preserving the remaining video features to retain broader contextual information. Furthermore, we design a region-aware cross-modal attention mechanism to enhance semantic alignment during translation. Experiments on a large-scale documentary translation dataset demonstrate that our method significantly outperforms baseline models, highlighting its effectiveness in long-video scenarios.

We present a solver-agnostic framework in which coordinated large language model (LLM) agents autonomously execute the complete computational mechanics workflow, from perceptual data of an engineering component through geometry extraction, material inference, discretisation, solver execution, uncertainty quantification, and code-compliant assessment, to an engineering report with actionable recommendations. Agents are formalised as conditioned operators on a shared context space with quality gates that introduce conditional iteration between pipeline layers. We introduce a mathematical framework for extracting engineering information from perceptual data under uncertainty using interval bounds, probability densities, and fuzzy membership functions, and introduce task-dependent conservatism to resolve the ambiguity of what `conservative' means when different limit states are governed by opposing parameter trends. The framework is demonstrated through a finite element analysis pipeline applied to a photograph of a steel L-bracket, producing a 171,504-node tetrahedral mesh, seven analyses across three boundary condition hypotheses, and a code-compliant assessment revealing structural failure with a quantified redesign. All results are presented as generated in the first autonomous iteration without manual correction, reinforcing that a professional engineer must review and sign off on any such analysis.

Large reasoning models (LRMs) have achieved remarkable performance in complex reasoning tasks, driven by their powerful inference-time scaling capability. However, LRMs often suffer from overthinking, which results in substantial computational redundancy and significantly reduces efficiency. Early-exit methods aim to mitigate this issue by terminating reasoning once sufficient evidence has been generated, yet existing approaches mostly rely on handcrafted or empirical indicators that are unreliable and impractical. In this work, we introduce Dynamic Thought Sufficiency in Reasoning (DTSR), a novel framework for efficient reasoning that enables the model to dynamically assess the sufficiency of its chain-of-thought (CoT) and determine the optimal point for early exit. Inspired by human metacognition, DTSR operates in two stages: (1) Reflection Signal Monitoring, which identifies reflection signals as potential cues for early exit, and (2) Thought Sufficiency Check, which evaluates whether the current CoT is sufficient to derive the final answer. Experimental results on the Qwen3 models show that DTSR reduces reasoning length by 28.9%-34.9% with minimal performance loss, effectively mitigating overthinking. We further discuss overconfidence in LRMs and self-evaluation paradigms, providing valuable insights for early-exit reasoning.

Previous research on multi-party dialogue generation has predominantly leveraged structural information inherent in dialogues to directly inform the generation process. However, the prevalence of colloquial expressions and incomplete utterances in dialogues often impedes comprehension and weakens the fidelity of dialogue structure representations, which is particularly pronounced in multi-party dialogues. In this work, we propose a novel framework DRCR (Discourse coherence and Response-guided Context Rewriting) to improve multi-party dialogue generation through dialogue context rewriting. Specifically, DRCR employs two complementary feedback signals, discourse coherence and response quality, to construct preference data for both context rewriting and response generation. Moreover, we propose a dynamic self-evolution learning method that allows the rewriter and responder to continuously enhance their capabilities through mutual interaction in an iterative training loop. Comprehensive experiments conducted on four multi-party dialogue datasets substantiate the effectiveness of DRCR.

We introduce Fleming-Viot Diffusion (FVD), an inference-time alignment method that resolves the diversity collapse commonly observed in Sequential Monte Carlo (SMC) based diffusion samplers. Existing SMC-based diffusion samplers often rely on multinomial resampling or closely related resampling schemes, which can still reduce diversity and lead to lineage collapse under strong selection pressure. Inspired by Fleming-Viot population dynamics, FVD replaces multinomial resampling with a specialized birth-death mechanism designed for diffusion alignment. To handle cases where rewards are only approximately available and naive rebirth would collapse deterministic trajectories, FVD integrates independent reward-based survival decisions with stochastic rebirth noise. This yields flexible population dynamics that preserve broader trajectory support while effectively exploring reward-tilted distributions, all without requiring value function approximation or costly rollouts. FVD is fully parallelizable and scales efficiently with inference compute. Empirically, it achieves substantial gains across settings: on DrawBench it outperforms prior methods by 7% in ImageReward, while on class-conditional tasks it improves FID by roughly 14-20% over strong baselines and is up to 66 times faster than value-based approaches.

Deep neural networks have emerged as powerful tools for learning operators defined over infinite-dimensional function spaces. However, existing theories frequently encounter difficulties related to dimensionality and limited interpretability. This work investigates how sparsity can help address these challenges in functional learning, a central ingredient in operator learning. We propose a framework that employs convolutional architectures to extract sparse features from a finite number of samples, together with deep fully connected networks to effectively approximate nonlinear functionals. Using universal discretization methods, we show that sparse approximators enable stable recovery from discrete samples. In addition, both the deterministic and the random sampling schemes are sufficient for our analysis. These findings lead to improved approximation rates and reduced sample sizes in various function spaces, including those with fast frequency decay and mixed smoothness. They also provide new theoretical insights into how sparsity can alleviate the curse of dimensionality in functional learning.

Conversational Query Rewriting (CQR) aims to rewrite ambiguous queries to achieve more efficient conversational search. Early studies have predominantly focused on the rewriting in isolation, ignoring the feedback from query rewrite, passage retrieval and response generation in the rewriting process. To address this issue, we propose Multi-Faceted Self-Consistent Preference Aligned CQR (MSPA-CQR). Specifically, we first construct self-consistent preference alignment data from three dimensions (rewriting, retrieval, and response) to generate more diverse rewritten queries. Then we propose prefix guided multi-faceted direct preference optimization to learn preference information from three different dimensions. The experimental results show that our MSPA-CQR is effective in both in- and out-of-distribution scenarios.

Maintenance procedures in manufacturing facilities are often documented as flowcharts in static PDFs or scanned images. They encode procedural knowledge essential for asset lifecycle management, yet inaccessible to modern operator support systems. Vision-language models, the dominant paradigm for image understanding, struggle to reconstruct connection topology from such diagrams. We present FlowExtract, a pipeline for extracting directed graphs from ISO 5807-standardized flowcharts. The system separates element detection from connectivity reconstruction, using YOLOv8 and EasyOCR for standard domain-aligned node detection and text extraction, combined with a novel edge detection method that analyzes arrowhead orientations and traces connecting lines backward to source nodes. Evaluated on industrial troubleshooting guides, FlowExtract achieves very high node detection and substantially outperforms vision-language model baselines on edge extraction, offering organizations a practical path toward queryable procedural knowledge representations. The implementation is available athttps://github.com/guille-gil/FlowExtract.

Language models operate on discrete tokens but compute in continuous vector spaces, inducing a Voronoi tessellation over the representation manifold. We study this tessellation empirically on Qwen3.5-4B-Base, making two contributions. First, using float32 margin recomputation to resolve bfloat16 quantization artifacts, we validate Mabrok's (2026) linear scaling law of the expressibility gap with $R^2$ = 0.9997 - the strongest confirmation to date - and identify a mid-layer geometric ambiguity regime where margin geometry is anti-correlated with cross-entropy (layers 24-28, $ρ$ = -0.29) before crystallizing into alignment at the final layer ($ρ$ = 0.836). Second, we show that the Voronoi tessellation of a converged model is reshapable through margin refinement procedures (MRP): short post-hoc optimization runs that widen token-decision margins without retraining. We compare direct margin maximization against Fisher information distance maximization across a dose-response sweep. Both methods find the same ceiling of ~16,300 correctable positions per 256K evaluated, but differ critically in collateral damage. Margin maximization damage escalates with intervention strength until corrections are overwhelmed. Fisher damage remains constant at ~5,300 positions across the validated range ($λ$ = 0.15-0.6), achieving +28% median margin improvement at $λ$ = 0.6 with invariant downstream benchmarks - a geometric reorganization that compresses the expressibility gap while preserving its scaling law. However, frequency and token-class audits reveal that gains concentrate in high-frequency structural tokens (84% of net corrections at $λ$ = 0.6), with content and entity-like contributions shrinking at higher $λ$. Fisher MRP is therefore a viable geometric polishing tool whose practical ceiling is set not by aggregate damage but by the uniformity of token-level benefit.

Recently, multi-Large Language Model (LLM) frameworks have been proposed to solve contextualized tasks. However, these frameworks do not explicitly emulate human team role division, which may lead to a single perspective, thereby weakening performance on multi-step contextualized tasks. To address this issue, we propose TeamLLM, a human-like Team-Oriented Multi-LLM Collaboration Framework. TeamLLM adopts four team roles with distinct division and employs a three-phase multi-LLM collaboration for multi-step contextualized tasks. To evaluate the effectiveness of TeamLLM on multi-step contextualized tasks, we propose Contextually-Grounded and Procedurally-Structured tasks (CGPST) and construct the CGPST benchmark. This benchmark has four core features: contextual grounding, procedural structure, process-oriented evaluation and multi-dimensional assessment. We evaluate ten popular LLMs on CGPST at overall-level, step-level, and dimension-level. Results show that TeamLLM substantially improves performance on CGPST. We release the benchmark with scenarios, full-process responses and human scores from ten LLMs. The code and data are available at https://anonymous.4open.science/r/TeamLLM-anonymous-C50E/.

Random GUI testing is a widely-used technique for testing mobile apps. However, its effectiveness is limited by the notorious issue -- UI exploration tarpits, where the exploration is trapped in local UI regions, thus impeding test coverage and bug discovery. In this experience paper, we introduce LLM-powered random GUI Testing, a novel hybrid testing approach to mitigating UI tarpits during random testing. Our approach monitors UI similarity to identify tarpits and query LLMs to suggest promising events for escaping the encountered tarpits. We implement our approach on top of two different automated input generation (AIG) tools for mobile apps: (1) HybridMonkey upon Monkey, a state-of-the-practice tool; and (2) HybridDroidbot upon Droidbot, a state-of-the-art tool. We evaluated them on 12 popular, real-world apps. The results show that HybridMonkey and HybridDroidbot outperform all baselines, achieving average coverage improvements of 54.8% and 44.8%, respectively, and detecting the highest number of unique crashes. In total, we found 75 unique bugs, including 34 previously unknown bugs. To date, 26 bugs have been confirmed and fixed. We also applied HybridMonkey on WeChat, a popular industrial app with billions of monthly active users. HybridMonkey achieved higher activity coverage and found more bugs than random testing.

We evaluate cognitive impairment (CI) classification from transcripts of speech in English, Slovene, and Korean. We compare zero-shot large language models (LLMs) used as direct classifiers under three input settings -- transcript-only, linguistic-features-only, and combined -- with supervised tabular approaches trained under a leave-one-out protocol. The tabular models operate on engineered linguistic features, transcript embeddings, and early or late fusion of both modalities. Across languages, zero-shot LLMs provide competitive no-training baselines, but supervised tabular models generally perform better, particularly when engineered linguistic features are included and combined with embeddings. Few-shot experiments focusing on embeddings indicate that the value of limited supervision is language-dependent, with some languages benefiting substantially from additional labelled examples while others remain constrained without richer feature representations. Overall, the results suggest that, in small-data CI detection, structured linguistic signals and simple fusion-based classifiers remain strong and reliable signals.

Large language models (LLMs) has been widely adopted as a scalable surrogate for human evaluation, yet such judges remain imperfect and susceptible to surface-level biases. One possible reason is that these judges lack sufficient information in assessing answer correctness. With the rise of reasoning-capable models, exposing a generator's reasoning content to the judge provides richer information and is a natural candidate for improving judgment accuracy. However, its actual impact on judge behavior remains understudied. In this paper, we systematically investigate how access to reasoning chains affects LLM-based judgment across factual question answering (QA) and mathematical reasoning benchmarks. We find that weak judges are easily swayed by reasoning presence, frequently accepting incorrect answers accompanied by fluent reasoning, while strong judges can partially leverage reasoning as informative evidence. Nevertheless, even strong judges are misled by seemingly high-quality reasoning chains. Controlled experiments further reveal that both fluency and factuality of reasoning chains are critical signals driving judge decisions. These findings highlight the need for more robust LLM judges that can distinguish genuine reasoning quality from superficial fluency when evaluating modern reasoning models.

Context: Large Language Models (LLMs) are increasingly used in modern software development, aiding in code generation, code completion, and refactoring through AI-powered assistants. While they accelerate development workflows, they often produce extraneous output, referred to as "babbling", which incurs additional cognitive, economic, and energy costs. Objective: This work investigates the problem of babbling in LLM-based code generation and proposes a practical, model-agnostic approach to reduce unnecessary output without compromising solution accuracy. Method: We introduce Babbling Suppression (BS), a method that integrates test execution into the LLM generation process by evaluating intermediate outputs and terminating generation once a solution passes all tests. This prevents excessive token generation while having no impact on model accuracy. An empirical study was conducted across two Python and two Java benchmarks, targeting four 3-4B parameter models and six 6-7B parameter models. Results: Our findings show that babbling occurs across all tested models, with higher frequency in Java than in Python. Applying BS significantly reduces energy consumption by up to 65% for Python and 62% for Java in models prone to babbling. Across 40 model-benchmark pairs, 29 showed reduced mean energy consumption, with reductions exceeding 20% in 22 cases. Generated token count decreased in 35 pairs, while the GPU energy-per-token overhead of BS remained below 10% for 26 pairs, decreased for 2, and reached a maximum of 24%, yielding net energy savings in most cases. Implications: BS can make AI-assisted programming more efficient and sustainable by reducing energy consumption and minimizing cognitive effort by developers. Its model-agnostic design allows easy integration, suggesting broad applicability.

I examine the technology of machine learning from the perspective of rhetoric, which is simply the art of persuasion. Rather than being a neutral and "objective" way to build "world models" from data, machine learning is (I argue) inherently rhetorical. I explore some of its rhetorical features, and examine one pervasive business model where machine learning is widely used, "manipulation as a service."

When an LLM-based agent improves on a task, is the gain from the model itself or from the reasoning paradigm wrapped around it? We study this question by comparing six inference-time paradigms, namely Direct, CoT, ReAct, Plan-Execute, Reflection, and ReCode, across four frontier LLMs and ten benchmarks, yielding roughly 18,000 runs. We find that reasoning structure helps dramatically on some tasks but hurts on others: ReAct improves over Direct by 44pp on GAIA, while CoT degrades performance by 15pp on HumanEval. No single paradigm dominates, and oracle per-task selection beats the best fixed paradigm by 17.1pp on average. Motivated by this complementarity, we propose a select-then-solve approach: before answering each task, a lightweight embedding-based router selects the most suitable paradigm. Across four models, the router improves average accuracy from 47.6% to 53.1%, outperforming the best fixed paradigm at 50.3% by 2.8pp and recovering up to 37% of the oracle gap. In contrast, zero-shot self-routing only works for GPT-5 at 67.1% and fails for weaker models, all trailing the learned router. Our results argue that reasoning paradigm selection should be a per-task decision made by a learned router, not a fixed architectural choice.

We present EmBolic - a novel fully hyperbolic deep learning architecture for fine-grained emotion analysis from textual messages. The underlying idea is that hyperbolic geometry efficiently captures hierarchies between both words and emotions. In our context, these hierarchical relationships arise from semantic ambiguities. EmBolic aims to infer the curvature on the continuous space of emotions, rather than treating them as a categorical set without any metric structure. In the heart of our architecture is the attention mechanism in the hyperbolic disc. The model is trained to generate queries (points in the hyperbolic disc) from textual messages, while keys (points at the boundary) emerge automatically from the generated queries. Predictions are based on the Busemann energy between queries and keys, evaluating how well a certain textual message aligns with the class directions representing emotions. Our experiments demonstrate strong generalization properties and reasonably good prediction accuracy even for small dimensions of the representation space. Overall, this study supports our claim that affective computing is one of the application domains where hyperbolic representations are particularly advantageous.

The aerodynamic design of turbomachinery is a complex and tightly coupled multi-stage process involving geometry generation, performance prediction, optimization, and high-fidelity physical validation. Existing intelligent design approaches typically focus on individual stages or rely on loosely coupled pipelines, making fully autonomous end-to-end design challenging.To address this issue, this study proposes TurboAgent, a large language model (LLM)-driven autonomous multi-agent framework for turbomachinery aerodynamic design and optimization. The LLM serves as the core for task planning and coordination, while specialized agents handle generative design, rapid performance prediction, multi-objective optimization, and physics-based validation. The framework transforms traditional trial-and-error design into a data-driven collaborative workflow, with high-fidelity simulations retained for final verification.A transonic single-rotor compressor is used for validation. The results show strong agreement between target performance, generated designs, and CFD simulations. The coefficients of determination (R2) for mass flow rate, total pressure ratio, and isentropic efficiency all exceed 0.91, with normalized RMSE values below 8%. The optimization agent further improves isentropic efficiency by 1.61% and total pressure ratio by 3.02%. The complete workflow can be executed within approximately 30 minutes under parallel computing. These results demonstrate that TurboAgent enables an autonomous closed-loop design process from natural language requirements to final design generation, providing an efficient and scalable paradigm for turbomachinery aerodynamic design

As Large Language Models (LLMs) scale to support context windows exceeding one million tokens, the linear growth of Key-Value (KV) cache imposes severe memory capacity and bandwidth bottlenecks, constraining the efficiency of long-context inference. Existing compression approaches typically prioritize tokens based on local saliency metrics to decouple prefill computation from decoding memory. However, these methods often rely on local saliency snapshots at a specific layer, thereby systematically discarding tokens that act as global information hubs across the network depth but appear temporarily dormant at the specific layer selected for pruning. To address this limitation, we propose StructKV, a structure-aware KV cache compression framework that introduces three core innovations: First, Global In-Degree Centrality aggregates attention patterns across the network depth to identify global information hubs. Second, Dynamic Pivot Detection utilizes information-theoretic metrics to adaptively locate the optimal layer for compression. Finally, Structural Propagation and Decoupling separates the computational budget from the memory storage budget. Experimental results on the LongBench and RULER benchmarks demonstrate that StructKV effectively preserves long-range dependencies and retrieval robustness.

Large Language Models (LLMs) are driving a shift towards intent-driven development, where agents build complete software from scratch. However, existing benchmarks fail to assess this 0-to-1 generation capability due to two limitations: reliance on predefined scaffolds that ignore repository structure planning, and rigid white-box unit testing that lacks end-to-end behavioral validation. To bridge this gap, we introduce CLI-Tool-Bench, a structure-agnostic benchmark for evaluating the ground-up generation of Command-Line Interface (CLI) tools. It features 100 diverse real-world repositories evaluated via a black-box differential testing framework. Agent-generated software is executed in sandboxes, comparing system side effects and terminal outputs against human-written oracles using multi-tiered equivalence metrics. Evaluating seven state-of-the-art LLMs, we reveal that top models achieve under 43% success, highlighting the ongoing challenge of 0-to-1 generation. Furthermore, higher token consumption does not guarantee better performance, and agents tend to generate monolithic code.

We study offline learning in KL-regularized two-player zero-sum games, where policies are optimized under a KL constraint to a fixed reference policy. Prior work relies on pessimistic value estimation to handle distribution shift, yielding only $\widetilde{\mathcal{O}}(1/\sqrt n)$ statistical rates. We develop a new pessimism-free algorithm and analytical framework for KL-regularized games, built on the smoothness of KL-regularized best responses and a stability property of the Nash equilibrium induced by skew symmetry. This yields the first $\widetilde{\mathcal{O}}(1/n)$ sample complexity bound for offline learning in KL-regularized zero-sum games, achieved entirely without pessimism. We further propose an efficient self-play policy optimization algorithm and prove that, with a number of iterations linear in the sample size, it achieves the same fast $\widetilde{\mathcal{O}}(1/n)$ statistical rate as the minimax estimator.

Large language models have demonstrated remarkable capabilities across a wide range of natural language processing tasks, yet their application in the legal domain remains challenging due to the specialized terminology, complex reasoning requirements, and rapidly evolving legal knowledge involved. In this paper, we present Luwen, an open-source Chinese legal language model built upon the Baichuan foundation model through three key techniques: continual pre-training on a large-scale legal corpus, supervised fine-tuning with carefully curated legal instruction data, and retrieval-augmented generation integrated with a comprehensive legal knowledge base. We evaluate Luwen on five representative legal tasks spanning both prediction and generation settings, including legal judgment prediction, judicial examination, legal text summarization, law article question answering, and judicial decision reasoning. Experimental results show that Luwen outperforms several strong baselines, demonstrating the effectiveness of our approach in adapting general-purpose language models to the legal domain.

Despite strong performance on Text-to-SQL benchmarks, it remains unclear whether LLM-generated SQL programs are structurally reliable. In this work, we investigate the structural behavior of LLM-generated SQL queries and introduce SQLStructEval, a framework for analyzing program structures through canonical abstract syntax tree (AST) representations. Our experiments on the Spider benchmark show that modern LLMs often produce structurally diverse queries for the same input, even when execution results are correct, and that such variance is frequently triggered by surface-level input changes such as paraphrases or schema presentation. We further show that generating queries in a structured space via a compile-style pipeline can improve both execution accuracy and structural consistency. These findings suggest that structural reliability is a critical yet overlooked dimension for evaluating LLM-based program generation systems. Our code is available at https://anonymous.4open.science/r/StructEval-2435.

Trial-and-error is a fundamental strategy for humans to solve complex problems and a necessary capability for Artificial Intelligence (AI) systems operating in real-world environments. Although several trial-and-error AI techniques have recently been proposed, most of them rely on simple heuristics designed by researchers and achieve limited performance gains. The core issue is the absence of appropriate data: current models cannot learn from detailed records of how humans actually conduct trial-and-error in practice. To address this gap, we introduce a data annotation platform and a corresponding dataset, termed Trial-and-Error Collection (TEC). The platform records users' complete trajectories across multiple trials and collects their reflections after receiving error feedback. Using this platform, we record the problem-solving processes of 46 participants on 58 tasks, resulting in 5,370 trial trajectories along with error reflections across 41,229 webpages. With this dataset, we observe that humans achieve substantially higher accuracy compared to LLMs, which demonstrates that humans are more effective in trial-and-error than LLMs. We believe that the TEC platform and dataset provide a valuable foundation for understanding human trial-and-error behavior and for developing more capable AI systems. Platform and dataset are publicly available.

Recent developments in hardware, such as photonic integrated circuits and optical devices, are driving demand for research on constructing machine learning architectures tailored for linear operations. Hence, it is valuable to explore methods for constructing learning machines with only linear operations after simple nonlinear preprocessing. In this study, we propose a framework to extract a linearized model from a pre-trained neural network for classification tasks by integrating Koopman operator theory with knowledge distillation. Numerical demonstrations on the MNIST and the Fashion-MNIST datasets reveal that the proposed model consistently outperforms the conventional least-squares-based Koopman approximation in both classification accuracy and numerical stability.

Multimodal sarcasm detection (MSD) aims to identify sarcastic intent from semantic incongruity between text and image. Although recent methods have improved MSD through cross-modal interaction and incongruity reasoning, they often assume that all modalities are equally reliable. In real-world social media, however, textual content may be ambiguous and visual content may be weakly relevant or even irrelevant, causing deterministic fusion to introduce noisy evidence and weaken robust reasoning. To address this issue, we propose Uncertainty-aware Robust Multimodal Fusion (URMF), a unified framework that explicitly models modality reliability during interaction and fusion. URMF first employs multi-head cross-attention to inject visual evidence into textual representations, followed by multi-head self-attention in the fused semantic space to enhance incongruity-aware reasoning. It then performs unified unimodal aleatoric uncertainty modeling over text, image, and interaction-aware latent representations by parameterizing each modality as a learnable Gaussian posterior. The estimated uncertainty is further used to dynamically regulate modality contributions during fusion, suppressing unreliable modalities and yielding a more robust joint representation. In addition, we design a joint training objective integrating task supervision, modality prior regularization, cross-modal distribution alignment, and uncertainty-driven self-sampling contrastive learning. Experiments on public MSD benchmarks show that URMF consistently outperforms strong unimodal, multimodal, and MLLM-based baselines, demonstrating the effectiveness of uncertainty-aware fusion for improving both accuracy and robustness.

Heterogeneity in time series data is more pronounced than in vision or language, as temporal dynamics vary substantially across domains and tasks. Existing efforts on training time series foundation models (TSFMs) from scratch are often trained with mixed-batch strategies that merge large-scale datasets, which can cause gradient conflicts and degrade representation quality. To address this, we propose a fine-grained learning method that distills invariant knowledge from heterogeneous series while reducing cross-domain interference. We characterize heterogeneity at two levels: inter-domain and intra-domain. To tackle this bi-level heterogeneity, we design a federated learning method that mitigates intra-domain conflicts by enforcing domain-invariant and semantically consistent representations through local regularization, and addresses inter-domain discrepancies by enhancing cross-domain collaboration via domain-aware aggregation. Experiments across diverse benchmarks show that TSFMs trained with our method consistently outperform both centralized and federated TSFM baselines in point and probabilistic forecasting, while also achieving competitive zero-shot performance at scale, offering a flexible pathway for training TSFMs from scratch in heterogeneous environments.

While traditional optimization problems were often studied in isolation, many real-world problems today require interdependence among multiple optimization components. The traveling thief problem (TTP) is a multi-component problem that has been widely studied in the literature. In this paper, we introduce and investigate the TTP with time window constraints which provides a TTP variant highly relevant to real-world situations where good can only be collected at given time intervals. We examine adaptions of existing approaches for TTP and the Traveling Salesperson Problem (TSP) with time windows to this new problem and evaluate their performance. Furthermore, we provide a new heuristic approach for the TTP with time windows. To evaluate algorithms for TTP with time windows, we introduce new TTP benchmark instances with time windows based on TTP instances existing in the literature. Our experimental investigations evaluate the different approaches and show that the newly designed algorithm outperforms the other approaches on a wide range of benchmark instances.

In today's AI-assisted software engineering landscape, developers increasingly depend on LLMs that are highly capable, yet inherently imperfect. The tendency of these models to produce incorrect outputs can reduce developer productivity. To this end, a canonical mitigation method is to provide calibrated confidence scores that faithfully reflect their likelihood of correctness at the instance-level. Such information allows users to make immediate decisions regarding output acceptance, abstain error-prone outputs, and better align their expectations with the model's capabilities. Since post-trained LLMs do not inherently produce well-calibrated confidence scores, researchers have developed post-hoc calibration methods, with global Platt-scaling of sequence-level confidence scores proving effective in many generative software engineering tasks but remaining unreliable or unexplored for automated code revision (ACR) tasks such as program repair, vulnerability repair, and code refinement. We hypothesise that the coarse-grained nature of this conventional method makes it ill-suited for ACR tasks, where correctness is often determined by local edit decisions and miscalibration can be sample-dependent, thereby motivating fine-grained confidence calibration. To address this, our study proposes local Platt-scaling applied separately to three different fine-grained confidence scores. Through experiments across 3 separate tasks and correctness metrics, as well as 14 different models of various sizes, we find that fine-grained confidence scores consistently achieve lower calibration error across a broader range of probability intervals, and this effect is further amplified when global Platt-scaling is applied. Our proposed approaches offer a practical solution to eliciting well-calibrated confidence scores, enabling more trustworthy and streamlined usage of imperfect models in ACR tasks.

Repurchase behavior is a primary signal in large-scale retail recommendation, particularly in categories with frequent replenishment: many items in a user's next basket were previously purchased and their timing follows stable, item-specific cadences. Yet most next basket repurchase recommendation models represent history as a sequence of discrete basket events indexed by visit order, which cannot explicitly model elapsed calendar time or update item rankings as days pass between purchases. We present CASE (Cadence-Aware Set Encoding for next basket repurchase recommendation), which decouples item-level cadence learning from cross-item interaction, enabling explicit calendar-time modeling while remaining production-scalable. CASE represents each item's purchase history as a calendar-time signal over a fixed horizon, applies shared multi-scale temporal convolutions to capture recurring rhythms, and uses induced set attention to model cross-item dependencies with sub-quadratic complexity, allowing efficient batch inference at scale. Across three public benchmarks and a proprietary dataset, CASE consistently improves Precision, Recall, and NDCG at multiple cutoffs compared to strong next basket prediction baselines. In a production-scale evaluation with tens of millions of users and a large item catalog, CASE achieves up to 8.6% relative Precision and 9.9% Recall lift at top-5, demonstrating that scalable cadence-aware modeling yields measurable gains in both benchmark and industrial settings.

Remote sensing semantic segmentation requires models that can jointly capture fine spatial details and high-level semantic context across complex scenes. While classical encoder-decoder architectures such as U-Net remain strong baselines, they often struggle to fully exploit global semantics and structured feature interactions. In this work, we propose HQF-Net, a hybrid quantum-classical multi-scale fusion network for remote sensing image segmentation. HQF-Net integrates multi-scale semantic guidance from a frozen DINOv3 ViT-L/16 backbone with a customized U-Net architecture through a Deformable Multiscale Cross-Attention Fusion (DMCAF) module. To enhance feature refinement, the framework further introduces quantum-enhanced skip connections (QSkip) and a Quantum bottleneck with Mixture-of-Experts (QMoE), which combines complementary local, global, and directional quantum circuits within an adaptive routing mechanism. Experiments on three remote sensing benchmarks show consistent improvements with the proposed design. HQF-Net achieves 0.8568 mIoU and 96.87% overall accuracy on LandCover.ai, 71.82% mIoU on OpenEarthMap, and 55.28% mIoU with 99.37% overall accuracy on SeasoNet. An architectural ablation study further confirms the contribution of each major component. These results show that structured hybrid quantum-classical feature processing is a promising direction for improving remote sensing semantic segmentation under near-term quantum constraints.

AI applications driven by multimodal large language models (MLLMs) are prone to hallucinations and pose considerable risks to human users. Crucially, such hallucinations are not equally problematic: some hallucination contents could be detected by human users(i.e., obvious hallucinations), while others are often missed or require more verification effort(i.e., elusive hallucinations). This indicates that multimodal AI hallucinations vary significantly in their verifiability. Yet, little research has explored how to control this property for AI applications with diverse security and usability demands. To address this gap, we construct a dataset from 4,470 human responses to AI-generated hallucinations and categorize these hallucinations into obvious and elusive types based on their verifiability by human users. Further, we propose an activation-space intervention method that learns separate probes for obvious and elusive hallucinations. We reveal that obvious and elusive hallucinations elicit different intervention probes, allowing for fine-grained control over the model's verifiability. Empirical results demonstrate the efficacy of this approach and show that targeted interventions yield superior performance in regulating corresponding verifiability. Moreover, simply mixing these interventions enables flexible control over the verifiability required for different scenarios.

Quantum computing simulators form the classical software foundation on which virtually all quantum algorithm research depends. We present Broken Quantum, the first comprehensive formal security audit of the open-source quantum computing simulator ecosystem. Applying COBALT QAI -- a four-module static analysis engine backed by the Z3 SMT solver -- we analyze 45 open-source quantum simulation frameworks from 22 organizations spanning 12 countries. We identify 547 security findings (40 CRITICAL, 492 HIGH, 15 MEDIUM) across four vulnerability classes: CWE-125/190 (C++ memory corruption), CWE-400 (Python resource exhaustion), CWE-502/94 (unsafe deserialization and code injection), and CWE-77/22 (QASM injection -- a novel, quantum-specific attack vector with no classical analog). All 13 vulnerability patterns are formally verified via Z3 satisfiability proofs (13/13 SAT). The 32-qubit boundary emerges as a consistent formal threshold in both C++ and Python vulnerability chains. Supply chain analysis identifies the first documented case of vulnerability transfer from a commercial quantum framework into US national laboratory infrastructure (IBM Qiskit Aer to XACC/Oak Ridge National Laboratory). Nine frameworks score 100/100 under all four scanners; Qiskit Aer,Cirq, tequila, PennyLane, and 5 others score 0/100.

Deciphering ancient Chinese Oracle Bone Script (OBS) is a challenging task that offers insights into the beliefs, systems, and culture of the ancient era. Existing approaches treat decipherment as a closed-set image recognition problem, which fails to bridge the ``interpretation gap'': while individual characters are often unique and rare, they are composed of a limited set of recurring, pictographic components that carry transferable semantic meanings. To leverage this structural logic, we propose an agent-driven Vision-Language Model (VLM) framework that integrates a VLM for precise visual grounding with an LLM-based agent to automate a reasoning chain of component identification, graph-based knowledge retrieval, and relationship inference for linguistically accurate interpretation. To support this, we also introduce OB-Radix, an expert-annotated dataset providing structural and semantic data absent from prior corpora, comprising 1,022 character images (934 unique characters) and 1,853 fine-grained component images across 478 distinct components with verified explanations. By evaluating our system across three benchmarks of different tasks, we demonstrate that our framework yields more detailed and precise decipherments compared to baseline methods.

We present ATANT (Automated Test for Acceptance of Narrative Truth), an open evaluation framework for measuring continuity in AI systems: the ability to persist, update, disambiguate, and reconstruct meaningful context across time. While the AI industry has produced memory components (RAG pipelines, vector databases, long context windows, profile layers), no published framework formally defines or measures whether these components produce genuine continuity. We define continuity as a system property with 7 required properties, introduce a 10-checkpoint evaluation methodology that operates without an LLM in the evaluation loop, and present a narrative test corpus of 250 stories comprising 1,835 verification questions across 6 life domains. We evaluate a reference implementation across 5 test suite iterations, progressing from 58% (legacy architecture) to 100% in isolated mode (250 stories) and 100% in 50-story cumulative mode, with 96% at 250-story cumulative scale. The cumulative result is the primary measure: when 250 distinct life narratives coexist in the same database, the system must retrieve the correct fact for the correct context without cross-contamination. ATANT is system-agnostic, model-independent, and designed as a sequenced methodology for building and validating continuity systems. The framework specification, example stories, and evaluation protocol are available at https://github.com/Kenotic-Labs/ATANT. The full 250-story corpus will be released incrementally.

Software systems are widely observed to grow in size, complexity, and interdependence over time, yet many large-scale systems remain stable despite persistent structural burden. This apparent tension suggests a limitation in one-dimensional views of software evolution. This paper introduces a graph-based, discrete-time probabilistic framework that separates structural burden from uncertainty. Change effort is modeled as a stochastic variable determined by the dependency neighborhood of the changed entity and by residual variability. Within this framework, burden is defined as expected effort and uncertainty as variance of effort. We show that, under explicit assumptions on non-decreasing average structural load, structural regularization, process stabilization, and covariance control, there exists a regime in which uncertainty decreases while structural burden does not. This regime formalizes the phenomenon of stabilization without simplification. The proposed framework provides a minimal theoretical explanation for how software systems can become more predictable over time without necessarily becoming structurally simpler, and offers a foundation for further theoretical and empirical studies of software evolution.

Autoencoders are widely used for dimensionality reduction, based on the assumption that high-dimensional data lies on low-dimensional manifolds. Regularized autoencoders aim to preserve manifold geometry during dimensionality reduction, but existing approaches often suffer from non-injective mappings and overly rigid constraints that limit their effectiveness and robustness. In this work, we identify encoder non-injectivity as a core bottleneck that leads to poor convergence and distorted latent representations. To ensure robustness across data distributions, we formalize the concept of admissible regularization and provide sufficient conditions for its satisfaction. In this work, we propose the Bi-Lipschitz Autoencoder (BLAE), which introduces two key innovations: (1) an injective regularization scheme based on a separation criterion to eliminate pathological local minima, and (2) a bi-Lipschitz relaxation that preserves geometry and exhibits robustness to data distribution drift. Empirical results on diverse datasets show that BLAE consistently outperforms existing methods in preserving manifold structure while remaining resilient to sampling sparsity and distribution shifts. Code is available at https://github.com/qipengz/BLAE.

Automated prompt optimization is crucial for eliciting reliable reasoning from large language models (LLMs), yet most API-only prompt optimizers iteratively edit monolithic prompts, coupling components and obscuring credit assignment, limiting controllability, and wasting tokens. We propose Adaptive Prompt Structure Factorization (aPSF), an API-only framework (prompt-in/text-out; no access to model internals) that uses an Architect model to discover task-specific prompt structures as semantic factors. aPSF then performs interventional, single-factor updates: interventional factor-level scoring estimates each factor's marginal contribution via validation-performance changes, and error-guided factor selection routes updates to the current dominant failure source for more sample-efficient optimization. Across multiple advanced reasoning benchmarks, aPSF outperforms strong baselines including principle-aware optimizers, improving accuracy by up to +2.16 percentage points on average, and reduces optimization cost by 45--87% tokens on MultiArith while reaching peak validation in 1 step.

The rapid development of AI agent systems is leading to an emerging Internet of Agents, where specialized agents operate across local devices, edge nodes, private services, and cloud platforms. Although recent efforts have improved agent naming, discovery, and interaction, efficient request dispatch remains an open systems problem under latency, privacy, and cost constraints. In this paper, we present AgentGate, a lightweight structured routing engine for candidate-aware agent dispatch. Instead of treating routing as unrestricted text generation, AgentGate formulates it as a constrained decision problem and decomposes it into two stages: action decision and structural grounding. The first stage determines whether a query should trigger single-agent invocation, multi-agent planning, direct response, or safe escalation, while the second stage instantiates the selected action into executable outputs such as target agents, structured arguments, or multi-step plans. To adapt compact models to this setting, we further develop a routing-oriented fine-tuning scheme with candidate-aware supervision and hard negative examples. Experiments on a curated routing benchmark with several 3B--7B open-weight models show that compact models can provide competitive routing performance in constrained settings, and that model differences are mainly reflected in action prediction, candidate selection, and structured grounding quality. These results indicate that structured routing is a feasible design point for efficient and privacy-aware agent systems, especially when routing decisions must be made under resource-constrained deployment conditions.

Large reasoning models (LRMs) that generate long chains of thought now perform well on multi-step math, science, and coding tasks. However, their behavior is still unstable and hard to interpret, and existing analysis tools struggle with such long, structured reasoning traces. We introduce Step-Saliency, which pools attention--gradient scores into step-to-step maps along the question--thinking--summary trajectory. Across several models, Step-Saliency reveals two recurring information-flow failures: Shallow Lock-in, where shallow layers over-focus on the current step and barely use earlier context, and Deep Decay, where deep layers gradually lose saliency on the thinking segment and the summary increasingly attends to itself and the last few steps. Motivated by these patterns, we propose StepFlow, a saliency-inspired test-time intervention that adjusts shallow saliency patterns measured by Step-Saliency via Odds-Equal Bridge and adds a small step-level residual in deep layers via Step Momentum Injection. StepFlow improves accuracy on math, science, and coding tasks across multiple LRMs without retraining, indicating that repairing information flow can recover part of their missing reasoning performance.

Real world deployment of multi agent reinforcement learning MARL systems is fundamentally constrained by limited compute memory and inference time. While expert policies achieve high performance they rely on costly decision cycles and large scale models that are impractical for edge devices or embedded platforms. Knowledge distillation KD offers a promising path toward resource aware execution but existing KD methods in MARL focus narrowly on action imitation often neglecting coordination structure and assuming uniform agent capabilities. We propose resource aware Knowledge Distillation for Multi Agent Reinforcement Learning KD MARL a two stage framework that transfers coordinated behavior from a centralized expert to lightweight decentralized student agents. The student policies are trained without a critic relying instead on distilled advantage signals and structured policy supervision to preserve coordination under heterogeneous and limited observations. Our approach transfers both action level behavior and structural coordination patterns from expert policies while supporting heterogeneous student architectures allowing each agent model capacity to match its observation complexity which is crucial for efficient execution under partial or limited observability and limited onboard resources. Extensive experiments on SMAC and MPE benchmarks demonstrate that KD MARL achieves high performance retention while substantially reducing computational cost. Across standard multi agent benchmarks KD MARL retains over 90 percent of expert performance while reducing computational cost by up to 28.6 times FLOPs. The proposed approach achieves expert level coordination and preserves it through structured distillation enabling practical MARL deployment across resource constrained onboard platforms.

Accurate classification requires not only high predictive accuracy but also well-calibrated confidence estimates. Yet, modern deep neural networks (DNNs) are often overconfident, primarily due to overfitting on the negative log-likelihood (NLL). While focal loss variants alleviate this issue, they typically reduce accuracy, revealing a persistent trade-off between calibration and predictive performance. Motivated by the complementary strengths of generative and discriminative classifiers, we propose Generative Cross-Entropy (GCE), which maximizes $p(x|y)$ and is equivalent to cross-entropy augmented with a class-level confidence regularizer. Under mild conditions, GCE is strictly proper. Across CIFAR-10/100, Tiny-ImageNet, and a medical imaging benchmark, GCE improves both accuracy and calibration over cross-entropy, especially in the long-tailed scenario. Combined with adaptive piecewise temperature scaling (ATS), GCE attains calibration competitive with focal-loss variants without sacrificing accuracy.

While Vision-Language Models (VLMs) have demonstrated significant potential in chemical visual understanding, current models are predominantly optimized for direct visual question-answering tasks. This paradigm often results in "black-box" systems that fail to utilize the inherent capability of Large Language Models (LLMs) to infer underlying reaction mechanisms. In this work, we introduce ChemVLR, a chemical VLM designed to prioritize reasoning within the perception process. Unlike conventional chemical VLMs, ChemVLR analyzes visual inputs in a fine-grained manner by explicitly identifying granular chemical descriptors, such as functional groups, prior to generating answers. This approach ensures the production of explicit and interpretable reasoning paths for complex visual chemical problems. To facilitate this methodology, we implement a cross-modality reverse-engineering strategy, combined with a rigorous filtering pipeline, to curate a large-scale reasoning-and-captioning dataset comprising 760k high-quality samples across molecular and reaction tasks. Furthermore, we adopt a three-stage training framework that systemically builds model perception and reasoning capacity. Experiments demonstrate that ChemVLR achieves state-of-the-art (SOTA) performance, surpassing both leading proprietary models and domain-specific open-source baselines. We also provide comprehensive ablation studies to validate our training strategy and data generation designs. Code and model weights will be available at https://github.com/xxlllz/ChemVLR.

Clinical Reasoning on Electronic Health Records (EHRs) is a fundamental yet challenging task in modern healthcare. While in-context learning (ICL) offers a promising inference-time adaptation paradigm for large language models (LLMs) in EHR reasoning, existing methods face three fundamental challenges: (1) Perspective Limitation, where data-driven similarity fails to align with LLM reasoning needs and model-driven signals are constrained by limited clinical competence; (2) Cohort Awareness, as demonstrations are selected independently without modeling population-level structure; and (3) Information Aggregation, where redundancy and interaction effects among demonstrations are ignored, leading to diminishing marginal gains. To address these challenges, we propose GraphWalker, a principled demonstration selection framework for EHR-oriented ICL. GraphWalker (i) jointly models patient clinical information and LLM-estimated information gain by integrating data-driven and model-driven perspectives, (ii) incorporates Cohort Discovery to avoid noisy local optima, and (iii) employs a Lazy Greedy Search with Frontier Expansion algorithm to mitigate diminishing marginal returns in information aggregation. Extensive experiments on multiple real-world EHR benchmarks demonstrate that GraphWalker consistently outperforms state-of-the-art ICL baselines, yielding substantial improvements in clinical reasoning performance. Our code is open-sourced at https://github.com/PuppyKnightUniversity/GraphWalker

Recently, Large Language Models (LLMs) have demonstrated significant potential in automating software engineering tasks. Generating software architecture designs from requirement documents is a crucial step in software development. However, there is currently a lack of functional datasets tailored for this task. To bridge this gap, we introduce R2ABench (Requirement-To-Architecture Benchmark), a novel benchmark comprising diverse real-world software projects paired with comprehensive Product Requirements Documents (PRDs) and expert-curated PlantUML reference diagrams. Furthermore, we propose a multi-dimensional, hybrid evaluation framework that assesses generated diagrams across three complementary layers: Structural Graph Metrics, Multi-dimensional Scoring, and Architecture Anti-pattern Detection. Using this framework, we conducted a comprehensive empirical study evaluating state-of-the-art models and agentic workflows. Our study shows that LLMs show strong syntactic validity and robust entity extraction but fundamentally struggle with relational reasoning, leading to structurally fragmented architectures. Code-specialized models partially alleviate this limitation, while agent frameworks introduce significant instability rather than consistent improvements. R2ABench provides a robust and standardized foundation for advancing LLM-driven software architecture generation.

Meaning in Persian poetry is both historical and relational. Words persist through literary tradition while shifting their force through changing constellations of neighbors, rhetorical frames, and poetic voices. This study examines that process using aligned Word2Vec spaces combined with graph-based neighborhood analysis across centuries and major poets. Rather than modeling semantic change as vector displacement alone, it treats lexical history as the rewiring of local semantic graphs: the gain and loss of neighbors, shifts in bridge roles, and movement across communities. The analysis centers on twenty target words, anchored by five recurrent reference terms: Earth, Night, two wine terms, and Heart. Surrounding them are affective, courtly, elemental, and Sufi concepts such as Love, Sorrow, Dervish, King, Annihilation, and Truth. These words exhibit distinct patterns of change. Night is more time-sensitive, Earth more poet-sensitive, and Heart shows continuity despite graph-role mobility. The two wine terms highlight probe sensitivity: one is broad and semantically diffuse, while the other is narrower and more stable. A lexical audit confirms that the corpus contains historically driven terms, poet-specific usages, and sparsely attested mystical vocabulary requiring caution. Overall, semantic change in Persian poetry is better captured as neighborhood rewiring than as abstract drift. For Digital Humanities, this approach restores local structure to computational analysis and supports interpretations closer to literary practice: persistence, migration, mediation, and selective transformation.

Explainable fake news detection aims to assess the veracity of news claims while providing human-friendly explanations. Existing methods incorporating investigative journalism are often inefficient and struggle with breaking news. Recent advances in large language models (LLMs) enable leveraging externally retrieved reports as evidence for detection and explanation generation, but unverified reports may introduce inaccuracies. Moreover, effective explainable fake news detection should provide a comprehensible explanation for all aspects of a claim to assist the public in verifying its accuracy. To address these challenges, we propose a graph-enhanced defense framework (G-Defense) that provides fine-grained explanations based solely on unverified reports. Specifically, we construct a claim-centered graph by decomposing the news claim into several sub-claims and modeling their dependency relationships. For each sub-claim, we use the retrieval-augmented generation (RAG) technique to retrieve salient evidence and generate competing explanations. We then introduce a defense-like inference module based on the graph to assess the overall veracity. Finally, we prompt an LLM to generate an intuitive explanation graph. Experimental results demonstrate that G-Defense achieves state-of-the-art performance in both veracity detection and the quality of its explanations.

Modern LLM service providers increasingly rely on autoscaling and parallelism reconfiguration to respond to rapidly changing workloads, but cold-start latency remains a major bottleneck. While recent systems have reduced model weight loading to seconds, CUDA graph capture still takes tens of seconds to minutes and often dominates startup. Unfortunately, CUDA graphs cannot be naively serialized: beyond graph topology, they are tightly coupled to execution context, including device addresses embedded in kernel arguments and kernel code lazily loaded during warmup. Existing approaches either rely on brittle kernel-specific patching or heavyweight process-level checkpoint/restore that are inflexible to dynamic parallelism switching. We present Foundry, a template-based CUDA graph context materialization system that persists both graph topology and execution context during an offline processing stage, and reconstructs executable graphs online with negligible overhead. Foundry enforces deterministic memory layouts, automatically extracts and reloads kernel binaries required by captured graphs, and reduces online reconstruction costs through topology-based templating. For distributed serving, Foundry further enables a single-GPU offline capture to generate templates for multi-GPU deployments by patching only rank-dependent communication state. Across dense and MoE models up to 235B parameters, Foundry reduces cold-start latency by up to 99%, cutting the initialization time of Qwen3-235B-A22B from 10 minutes to 3.9 seconds while preserving the throughput gains of CUDA graphs.

Large Language Models (LLMs) are increasingly used to simulate social attitudes and behaviors, offering scalable "silicon samples" that can approximate human data. However, current simulation practice often collapses diversity into an "average persona," masking subgroup variation that is central to social reality. This study introduces audience segmentation as a systematic approach for restoring heterogeneity in LLM-based social simulation. Using U.S. climate-opinion survey data, we compare six segmentation configurations across two open-weight LLMs (Llama 3.1-70B and Mixtral 8x22B), varying segmentation identifier granularity, parsimony, and selection logic (theory-driven, data-driven, and instrument-based). We evaluate simulation performance with a three-dimensional evaluation framework covering distributional, structural, and predictive fidelity. Results show that increasing identifier granularity does not produce consistent improvement: moderate enrichment can improve performance, but further expansion does not reliably help and can worsen structural and predictive fidelity. Across parsimony comparisons, compact configurations often match or outperform more comprehensive alternatives, especially in structural and predictive fidelity, while distributional fidelity remains metric dependent. Identifier selection logic determines which fidelity dimension benefits most: instrument-based selection best preserves distributional shape, whereas data-driven selection best recovers between-group structure and identifier-outcome associations. Overall, no single configuration dominates all dimensions, and performance gains in one dimension can coincide with losses in another. These findings position audience segmentation as a core methodological approach for valid LLM-based social simulation and highlight the need for heterogeneity-aware evaluation and variance-preserving modeling strategies.

Generated contents have raised serious concerns about copyright protection, image provenance, and credit attribution. A potential solution for these problems is watermarking. Recently, content watermarking for text-to-image diffusion models has been studied extensively for its effective detection utility and robustness. However, these watermarking techniques are vulnerable to potential adversarial attacks, such as removal attacks and forgery attacks. In this paper, we build a novel watermarking paradigm called Instance-Specific watermarking with Two-Sided detection (ISTS) to resist removal and forgery attacks. Specifically, we introduce a strategy that dynamically controls the injection time and watermarking patterns based on the semantics of users' prompts. Furthermore, we propose a new two-sided detection approach to enhance robustness in watermark detection. Experiments have demonstrated the superiority of our watermarking against removal and forgery attacks.

Adaptive moment methods such as Adam use a diagonal, coordinate-wise preconditioner based on exponential moving averages of squared gradients. This diagonal scaling is coordinate-system dependent and can struggle with dense or rotated parameter couplings, including those in matrix factorization, tensor decomposition, and graph neural networks, because it treats each parameter independently. We introduce FlowAdam, a hybrid optimizer that augments Adam with continuous gradient-flow integration via an ordinary differential equation (ODE). When EMA-based statistics detect landscape difficulty, FlowAdam switches to clipped ODE integration. Our central contribution is Soft Momentum Injection, which blends ODE velocity with Adam's momentum during mode transitions. This prevents the training collapse observed with naive hybrid approaches. Across coupled optimization benchmarks, the ODE integration provides implicit regularization, reducing held-out error by 10-22% on low-rank matrix/tensor recovery and 6% on Jester (real-world collaborative filtering), also surpassing tuned Lion and AdaBelief, while matching Adam on well-conditioned workloads (CIFAR-10). MovieLens-100K confirms benefits arise specifically from coupled parameter interactions rather than bias estimation. Ablation studies show that soft injection is essential, as hard replacement reduces accuracy from 100% to 82.5%.

Existing prompt-based fine-tuning methods typically learn task-specific prompts independently, imposing significant computing and storage overhead at scale when deploying multiple clinical natural language processing (NLP) systems. We present a multitask prompt distillation and decomposition framework that learns a single shared metaprompt from 21 diverse clinical source tasks and adapts it to unseen target tasks with fewer than 0.05% trainable parameters. Evaluated across five clinical NLP task types (named entity recognition, relation extraction, question answering, natural language inference, and summarization) on 10 held-out target datasets using three backbone models (LLaMA 3.1 8B, Meditron3 8B, gpt-oss 20B), our framework consistently outperforms LoRA by 1.5~1.7% despite using orders of magnitude fewer parameters, and exceeds single-task prompt tuning by 6.1~6.6%. The gpt-oss 20B model achieves the highest overall performance, particularly on clinical reasoning tasks. The strong zero- and few-shot performance demonstrates better transferability of the shared prompt representation.

Retrieval-Augmented Generation (RAG) systems are typically evaluated under static assumptions, despite being frequently corrected through user or expert feedback in deployment. Existing evaluation protocols focus on overall accuracy and fail to capture how systems adapt after feedback is introduced. We introduce feedback adaptation as a problem setting for RAG systems, which asks how effectively and how quickly corrective feedback propagates to future queries. To make this behavior measurable, we propose two evaluation axes: correction lag, which captures the delay between feedback provision and behavioral change, and post-feedback performance, which measures reliability on semantically related queries after feedback. Using these metrics, we show that training-based approaches exhibit a trade-off between delayed correction and reliable adaptation. We further propose PatchRAG, a minimal inference-time instantiation that incorporates feedback without retraining, demonstrating immediate correction and strong post-feedback generalization under the proposed evaluation. Our results highlight feedback adaptation as a previously overlooked dimension of RAG system behavior in interactive settings.

As deep learning inference is increasingly deployed in shared and cloud-based settings, a growing concern is input repurposing, in which data submitted for one task is reused by unauthorized models for another. Existing privacy defenses largely focus on restricting data access, but provide limited control over what downstream uses a released representation can still support. We propose a feature extraction framework that suppresses cross-model transfer while preserving accuracy for a designated classifier. The framework employs a variational latent bottleneck, trained with a task-driven cross-entropy objective and KL regularization, but without any pixel-level reconstruction loss, to encode inputs into a compact latent space. A dynamic binary mask, computed from per-dimension KL divergence and gradient-based saliency with respect to the frozen target model, suppresses latent dimensions that are uninformative for the intended task. Because saliency computation requires gradient access, the encoder is trained in a white-box setting, whereas inference requires only a forward pass through the frozen target model. On CIFAR-100, the processed representations retain strong utility for the designated classifier while reducing the accuracy of all unintended classifiers to below 2%, yielding a suppression ratio exceeding 45 times relative to unintended models. Preliminary experiments on CIFAR-10, Tiny ImageNet, and Pascal VOC provide exploratory evidence that the approach extends across task settings, although further evaluation is needed to assess robustness against adaptive adversaries.

Effective detection of unknown network security threats in multi-class imbalanced environments is critical for maintaining cyberspace security. Current methods focus on learning class representations but face challenges with unknown threat detection, class imbalance, and lack of interpretability, limiting their practical use. To address this, we propose RPM-Net, a novel framework that introduces reciprocal point mechanism to learn "non-class" representations for each known attack category, coupled with adversarial margin constraints that provide geometric interpretability for unknown threat detection. RPM-Net++ further enhances performance through Fisher discriminant regularization. Experimental results show that RPM-Net achieves superior performance across multiple metrics including F1-score, AUROC, and AUPR-OUT, significantly outperforming existing methods and offering practical value for real-world network security applications. Our code is available at:https://github.com/chiachen-chang/RPM-Net

Process supervision has emerged as a promising approach for enhancing LLM reasoning, yet existing methods fail to distinguish meaningful progress from mere verbosity, leading to limited reasoning capabilities and unresolved token inefficiency. To address this, we propose Stage-aware Hierarchical Advantage via Potential Estimation (SHAPE), a framework that formalizes reasoning as a trajectory through a state space of empirical solvability. SHAPE introduces a hierarchical credit assignment mechanism: at the segment level, it employs a stage-aware advantage function to prioritize efficient breakthroughs in low-potential states; at the token level, it utilizes entropy-driven redistribution to sharpen execution signals. Extensive experiments in math reasoning across three base models and five benchmarks demonstrate that SHAPE achieves an average accuracy gain of 3% with 30% reduced token consumption.

Recent advancements in Large Language Models (LLMs) have sparked interest in their application to Static Application Security Testing (SAST), primarily due to their superior contextual reasoning capabilities compared to traditional symbolic or rule-based methods. However, existing LLM-based approaches typically attempt to replace human experts directly without integrating effectively with existing SAST tools. This lack of integration results in ineffectiveness, including high rates of false positives, hallucinations, limited reasoning depth, and excessive token usage, making them impractical for industrial deployment. To overcome these limitations, we present a paradigm shift that reorchestrates the SAST workflow from current LLM-assisted structure to a new LLM-centered workflow. We introduce Argus (Agentic and Retrieval-Augmented Guarding System), the first multi-agent framework designed specifically for vulnerability detection. Argus incorporates three key novelties: comprehensive supply chain analysis, collaborative multi-agent workflows, and the integration of state-of-the-art techniques such as Retrieval-Augmented Generation (RAG) and ReAct to minimize hallucinations and enhance reasoning. Extensive empirical evaluation demonstrates that Argus significantly outperforms existing methods by detecting a higher volume of true vulnerabilities while simultaneously reducing false positives and operational costs. Notably, Argus has identified several critical zero-day vulnerabilities with CVE assignments.

Federated Learning (FL) enables collaborative model training while preserving data privacy, but its practical deployment is hampered by system and statistical heterogeneity. While federated network pruning offers a path to mitigate these issues, existing methods face a critical dilemma: server-side pruning lacks personalization, whereas client-side pruning is computationally prohibitive for resource-constrained devices. Furthermore, the pruning process itself induces significant parametric divergence among heterogeneous submodels, destabilizing training and hindering global convergence. To address these challenges, we propose SubFLOT, a novel framework for server-side personalized federated pruning. SubFLOT introduces an Optimal Transport-enhanced Pruning (OTP) module that treats historical client models as proxies for local data distributions, formulating the pruning task as a Wasserstein distance minimization problem to generate customized submodels without accessing raw data. Concurrently, to counteract parametric divergence, our Scaling-based Adaptive Regularization (SAR) module adaptively penalizes a submodel's deviation from the global model, with the penalty's strength scaled by the client's pruning rate. Comprehensive experiments demonstrate that SubFLOT consistently and substantially outperforms state-of-the-art methods, underscoring its potential for deploying efficient and personalized models on resource-constrained edge devices.

We present Logical Robots, an interactive multi-agent simulation platform where autonomous robot behavior is specified declaratively in the logic programming language Logica. Robot behavior is defined by logical predicates that map observations from simulated radar arrays and shared memory to desired motor outputs. This approach allows low-level reactive control and high-level planning to coexist within a single programming environment, providing a coherent framework for exploring multi-agent robot behavior.

A prevailing narrative in LLM post-training holds that supervised finetuning (SFT) memorizes while reinforcement learning (RL) generalizes. We revisit this claim for reasoning SFT with long chain-of-thought (CoT) supervision and find that cross-domain generalization is not absent but conditional, jointly shaped by optimization dynamics, training data, and base-model capability. Some reported failures are under-optimization artifacts: cross-domain performance first degrades before recovering and improving with extended training (a dip-and-recovery pattern), so shorttraining checkpoints can underestimate generalization. Data quality and structure both matter: low-quality solutions broadly hurt generalization,while verified long-CoT traces yield consistent cross-domain gains. Model capability is essential: stronger models internalize transferable procedural patterns (e.g., backtracking) even from a toy arithmetic game, while weaker ones imitate surface verbosity. This generalization is asymmetric, however: reasoning improves while safety degrades, reframing the question from whether reasoning SFT generalizes to under what conditions and at what cost.

In-Context Learning and Chain-of-Thought prompting improve reasoning in large language models (LLMs). These typically come at the cost of longer, more expensive prompts that may contain redundant information. Prompt compression based on pruning offers a practical solution, yet existing methods rely on sequential token removal which is computationally intensive. We present DiffuMask, a diffusion-based framework integrating hierarchical shot-level and token-level pruning signals, that enables rapid and parallel prompt pruning via iterative mask prediction. DiffuMask substantially accelerates the compression process via masking multiple tokens in each denoising step. It offers tunable control over retained content, preserving essential reasoning context and achieving up to 80\% prompt length reduction. Meanwhile, it maintains or improves accuracy across in-domain, out-of-domain, and cross-model settings. Our results show that DiffuMask provides a generalizable and controllable framework for prompt compression, facilitating faster and more reliable in-context reasoning in LLMs.

Dr. David Blackwell was a mathematician and statistician of the first rank, whose contributions to statistical theory, game theory, and decision theory predated many of the algorithmic breakthroughs that define modern artificial intelligence. This survey examines three of his most consequential theoretical results the Rao Blackwell theorem, the Blackwell Approachability theorem, and the Blackwell Informativeness theorem (comparison of experiments) and traces their direct influence on contemporary AI and machine learning. We show that these results, developed primarily in the 1940s and 1950s, remain technically live across modern subfields including Markov Chain Monte Carlo inference, autonomous mobile robot navigation (SLAM), generative model training, no-regret online learning, reinforcement learning from human feedback (RLHF), large language model alignment, and information design. NVIDIAs 2024 decision to name their flagship GPU architecture (Blackwell) provides vivid testament to his enduring relevance. We also document an emerging frontier: explicit Rao Blackwellized variance reduction in LLM RLHF pipelines, recently proposed but not yet standard practice. Together, Blackwell theorems form a unified framework addressing information compression, sequential decision making under uncertainty, and the comparison of information sources precisely the problems at the core of modern AI.

Quantitative estimation of wheel polygonal roughness from axle-box vibration signals is a challenging yet practically relevant problem for rail-vehicle condition monitoring. Existing studies have largely focused on detection, identification, or severity classification, while continuous regression of multi-order roughness spectra remains less explored, especially under real operational data and unseen-wheel conditions. To address this problem, this paper presents PD-SOVNet, a physics-guided gray-box framework that combines shared second-order vibration kernels, a $4\times4$ MIMO coupling module, an adaptive physical correction branch, and a Mamba-based temporal branch for estimating the 1st--40th-order wheel roughness spectrum from axle-box vibrations. The proposed design embeds modal-response priors into the model while retaining data-driven flexibility for sample-dependent correction and residual temporal dynamics. Experiments on three real-world datasets, including operational data and real fault data, show that the proposed method provides competitive prediction accuracy and relatively stable cross-wheel performance under the current data protocol, with its most noticeable advantage observed on the more challenging Dataset III. Noise injection experiments further indicate that the Mamba temporal branch helps mitigate performance degradation under perturbed inputs. These results suggest that structured physical priors can be beneficial for stabilizing roughness regression in practical rail-vehicle monitoring scenarios, although further validation under broader operating conditions and stricter comparison protocols is still needed.

Hybrid queries combining high-dimensional vector similarity search with spatio-temporal filters are increasingly critical for modern retrieval-augmented generation (RAG) systems. Existing systems typically handle these workloads by nesting vector indices within low-dimensional spatial structures, such as R-trees. However, this decoupled architecture fragments the vector space, forcing the query engine to invoke multiple disjoint sub-indices per query. This fragmentation destroys graph routing connectivity, incurs severe traversal overhead, and struggles to optimize for complex spatial boundaries. In this paper, we propose CubeGraph, a novel indexing framework designed to natively integrate vector search with arbitrary spatial constraints. CubeGraph partitions the spatial domain using a hierarchical grid, maintaining modular vector graphs within each cell. During query execution, CubeGraph dynamically stitches together adjacent cube-level indices on the fly whenever their spatial cells intersect with the query filter. This dynamic graph integration restores global connectivity, enabling a unified, single-pass nearest-neighbor traversal that eliminates the overhead of fragmented sub-index invocations. Extensive evaluations on real-world datasets demonstrate that CubeGraph significantly outperforms state-of-the-art baselines, offering superior query execution performance, scalability, and flexibility for complex hybrid workloads.

Prompt learning has gained significant attention as a parameter-efficient approach for adapting large pre-trained vision-language models to downstream tasks. However, when only partial labels are available, its performance is often limited by label ambiguity and insufficient supervisory information. To address this issue, we propose Holistic Optimal Label Selection (HopS), leveraging the generalization ability of pre-trained feature encoders through two complementary strategies. First, we design a local density-based filter that selects the top frequent labels from the nearest neighbors' candidate sets and uses the softmax scores to identify the most plausible label, capturing structural regularities in the feature space. Second, we introduce a global selection objective based on optimal transport that maps the uniform sampling distribution to the candidate label distributions across a batch. By minimizing the expected transport cost, it can determine the most likely label assignments. These two strategies work together to provide robust label selection from both local and global perspectives. Extensive experiments on eight benchmark datasets show that HopS consistently improves performance under partial supervision and outperforms all baselines. Those results highlight the merit of holistic label selection and offer a practical solution for prompt learning in weakly supervised settings.

Modern reasoning models continue generating long after the answer is already determined. Across five model configurations, two families, and three benchmarks, we find that \textbf{52--88\% of chain-of-thought tokens are produced after the answer is recoverable} from a partial prefix. This post-commitment generation reveals a structural phenomenon: the \textbf{detection--extraction gap}. Free continuations from early prefixes recover the correct answer even at 10\% of the trace, while forced extraction fails on 42\% of these cases. The answer is recoverable from the model state, yet prompt-conditioned decoding fails to extract it. We formalize this mismatch via a total-variation bound between free and forced continuation distributions, yielding quantitative estimates of suffix-induced shift. Exploiting this asymmetry, we propose Black-box Adaptive Early Exit (\BAEE{}), which uses free continuations for both detection and extraction, truncating \textbf{70--78\% of serial generation} while \textbf{improving accuracy by 1--5\,pp} across all models. For thinking-mode models, early exit prevents post-commitment overwriting, yielding gains of up to 5.8\,pp; a cost-optimized variant achieves 68--73\% reduction at a median of 9 API calls. Code is available at https://github.com/EdWangLoDaSc/know2say.

Shape optimisation of thin-shell structures requires a flexible, differentiable geometric representation suitable for gradient-based optimisation. We propose a neural parametric representation (NRep) for the shell mid-surface based on a neural network with periodic activation functions. The NRep is defined using a multi-layer perceptron (MLP), which maps the parametric coordinates of mid-surface vertices to their physical coordinates. A structural compliance optimisation problem is posed to optimise the shape of a thin-shell parameterised by the NRep subject to a volume constraint, with the network parameters as design variables. The resulting shape optimisation problem is solved using a gradient-based optimisation algorithm. Benchmark examples with classical solutions demonstrate the effectiveness of the proposed NRep. The approach exhibits potential for complex lattice-skin structures, owing to the compact and expressive geometry representation afforded by the NRep.

Decentralised online learning enables runtime adaptation in cyber-physical multi-agent systems, but when operating conditions change, learned policies often require substantial trial-and-error interaction before recovering performance. To address this, we propose TwinLoop, a simulation-in-the-loop digital twin framework for online multi-agent reinforcement learning. When a context shift occurs, the digital twin is triggered to reconstruct the current system state, initialise from the latest agent policies, and perform accelerated policy improvement with simulation what-if analysis before synchronising updated parameters back to the agents in the physical system. We evaluate TwinLoop in a vehicular edge computing task-offloading scenario with changing workload and infrastructure conditions. The results suggest that digital twins can improve post-shift adaptation efficiency and reduce reliance on costly online trial-and-error.

Large language models (LLMs) have shown strong knowledge reserves and task-solving capabilities, but still face the challenge of severe hallucination, hindering their practical application. Though scientific theories and rules can efficiently direct the behaviors of human manipulators, LLMs still do not utilize these highly-condensed knowledge sufficiently through training or prompting. To address this issue, we propose \textbf{SciDC}, an LLM generation method that integrate subject-specific knowledge with strong constraints. By adopting strong LLMs to automatically convert flexible knowledge into multi-layered, standardized rules, we build an extensible framework to effectively constrain the model generation on domain tasks. Experiments on scientific tasks including industrial formulation design, clinical tumor diagnosis and retrosynthesis planning, consistently demonstrate the effectiveness of our method, achieving a 12\% accuracy improvement on average compared with vanilla generation. We further discuss the potential of LLMs in automatically inductively summarizing highly-condensed knowledge, looking ahead to practical solutions for accelerating the overall scientific research process. All the code of this paper can be obtained (https://github.com/Maotian-Ma/SciDC).

Semi-supervised learning aims to infer class labels using only a small fraction of labeled data. In graph-based semi-supervised learning, this is typically achieved through label propagation to predict labels of unlabeled nodes. However, in real-world applications, data often arrive incrementally in batches. Each time a new batch appears, reapplying the traditional label propagation algorithm to recompute all labels is redundant, computationally intensive, and inefficient. To address the absence of an efficient label propagation update method, we propose DynLP, a novel GPU-centric Dynamic Batched Parallel Label Propagation algorithm that performs only the necessary updates, propagating changes to the relevant subgraph without requiring full recalculation. By exploiting GPU architectural optimizations, our algorithm achieves on average 13x and upto 102x speedup on large-scale datasets compared to state-of-the-art approaches.

Context: Corporate education is a strategic investment in the software industry, but little is known about how different professional profiles perceive these initiatives. Objective: To investigate whether sociodemographic and professional variables influence the perception of quality and effectiveness of corporate training in Software Engineering (SE). Method: Non-parametric significance tests were applied to data from a survey with 282 Brazilian professionals, crossing 27 perception items with 9 sociodemographic variables (gender, age, education level, state, experience, professional level, company size, area of work, and nature of participation), totaling 243 combinations. Results: Of the 243 combinations tested, only 35 showed statistical significance. Training mandatoriness was the dominant factor, affecting 24 of 27 items. Length of experience revealed a non-linear descriptive pattern with a low-engagement zone between 3 and 6 years. Differences by area of work indicated an expressive gap in soft skills training for advanced technical roles. Personal profile variables and company size produced no relevant significant differences. Conclusion: Personal profile variables do not determine the perception of quality and effectiveness, while professional trajectory variables (experience, level, area of work) produce localized differences. The voluntariness of participation remains a determining factor, in line with the literature. The absence of gender differences in a sample with 23\% women suggests that barriers operate before training, in access and representation, not during the learning experience.

A Grammatical Error Correction (GEC) system produces a sequence of edits to correct an erroneous sentence. The quality of these edits is typically evaluated against human annotations. However, a sentence may admit multiple valid corrections, and existing evaluation settings do not fully accommodate diverse application scenarios. Recent meta-evaluation approaches rely on human judgments across multiple references, but they are difficult to scale to large datasets. In this paper, we propose a new task, Scoring Edit Impact in GEC, which aims to automatically estimate the importance of edits produced by a GEC system. To address this task, we introduce a scoring framework based on an embedded association graph. The graph captures latent dependencies among edits and syntactically related edits, grouping them into coherent groups. We then perform perplexity-based scoring to estimate each edit's contribution to sentence fluency. Experiments across 4 GEC datasets, 4 languages, and 4 GEC systems demonstrate that our method consistently outperforms a range of baselines. Further analysis shows that the embedded association graph effectively captures cross-linguistic structural dependencies among edits.

Missing-person and child-safety investigations rely on heterogeneous case documents, including structured forms, bulletin-style posters, and narrative web profiles. Variations in layout, terminology, and data quality impede rapid triage, large-scale analysis, and search-planning workflows. This paper introduces the Guardian Parser Pack, an AI-driven parsing and normalization pipeline that transforms multi-source investigative documents into a unified, schema-compliant representation suitable for operational review and downstream spatial modeling. The proposed system integrates (i) multi-engine PDF text extraction with Optical Character Recognition (OCR) fallback, (ii) rule-based source identification with source-specific parsers, (iii) schema-first harmonization and validation, and (iv) an optional Large Language Model (LLM)-assisted extraction pathway incorporating validator-guided repair and shared geocoding services. We present the system architecture, key implementation decisions, and output design, and evaluate performance using both gold-aligned extraction metrics and corpus-level operational indicators. On a manually aligned subset of 75 cases, the LLM-assisted pathway achieved substantially higher extraction quality than the deterministic comparator (F1 = 0.8664 vs. 0.2578), while across 517 parsed records per pathway it also improved aggregate key-field completeness (96.97\% vs. 93.23\%). The deterministic pathway remained much faster (mean runtime 0.03 s/record vs. 3.95 s/record for the LLM pathway). In the evaluated run, all LLM outputs passed initial schema validation, so validator-guided repair functioned as a built-in safeguard rather than a contributor to the observed gains. These results support controlled use of probabilistic AI within a schema-first, auditable pipeline for high-stakes investigative settings.

As the complexity of modern workloads and hardware increasingly outpaces human research and engineering capacity, existing methods for database performance optimization struggle to keep pace. To address this gap, a new class of techniques, termed AI-Driven Research for Systems (ADRS), uses large language models to automate solution discovery. This approach shifts optimization from manual system design to automated code generation. The key obstacle, however, in applying ADRS is the evaluation pipeline. Since these frameworks rapidly generate hundreds of candidates without human supervision, they depend on fast and accurate feedback from evaluators to converge on effective solutions. Building such evaluators is especially difficult for complex database systems. To enable the practical application of ADRS in this domain, we propose automating the design of evaluators by co-evolving them with the solutions. We demonstrate the effectiveness of this approach through three case studies optimizing buffer management, query rewriting, and index selection. Our automated evaluators enable the discovery of novel algorithms that outperform state-of-the-art baselines (e.g., a deterministic query rewrite policy that achieves up to 6.8x lower latency), demonstrating that addressing the evaluation bottleneck unlocks the potential of ADRS to generate highly optimized, deployable code for next-generation data systems.

Small language models (SLM) are increasingly used as interactive decision-making agents, yet most decision-oriented evaluations ignore emotion as a causal factor influencing behavior. We study emotion-sensitive decision making by combining representation-level emotion induction with a structured game-theoretic evaluation. Emotional states are induced using activation steering derived from crowd-validated, real-world emotion-eliciting texts, enabling controlled and transferable interventions beyond prompt-based methods. We introduce a benchmark built around canonical decision templates that span cooperative and competitive incentives under both complete and incomplete information. These templates are instantiated using strategic scenarios from \textsc{Diplomacy}, \textsc{StarCraft II}, and diverse real-world personas. Experiments across multiple model families in various architecture and modalities, show that emotional perturbations systematically affect strategic choices, but the resulting behaviors are often unstable and not fully aligned with human expectations. Finally, we outline an approach to improve robustness to emotion-driven perturbations.

High-dimensional hyperspectral imaging (HSI) enables the visualization of ultrafast molecular dynamics and complex, heterogeneous spectra. However, applying this capability to resolve spatially varying vibrational couplings in two-dimensional infrared (2DIR) spectroscopy, a type of coherent multidimensional spectroscopy (CMDS), necessitates prohibitively long data acquisition, driven by dense Nyquist sampling requirements and the need for extensive signal accumulation. To address this challenge, we introduce a physics-informed neural representation approach that efficiently reconstructs dense spatially-resolved 2DIR hyperspectral images from sparse experimental measurements. In particular, we used a multilayer perceptron (MLP) to model the relationship between the sub-sampled 4D coordinates and their corresponding spectral intensities, and recover densely sampled 4D spectra from limited observations. The reconstruction results demonstrate that our method, using a fraction of the samples, faithfully recovers both oscillatory and non-oscillatory spectral dynamics in experimental measurement. Moreover, we develop a loss-aware adaptive sampling method to progressively introduce potentially informative samples for iterative data collection while conducting experiments. Experimental results show that the proposed approach achieves high-fidelity spectral recovery using only $1/32$ of the sampling budget, as opposed to exhaustive sampling, effectively reducing total experiment time by up to 32-fold. This framework offers a scalable solution for accelerating any experiments with hypercube data, including multidimensional spectroscopy and hyperspectral imaging, paving the way for rapid chemical imaging of transient biological and material systems.

Understanding and explaining the structure of generated test inputs is essential for effective software testing and debugging. Existing approaches--including grammar-based fuzzers, probabilistic Context-Free Grammars (pCFGs), and Large Language Models (LLMs)--suffer from critical limitations. They frequently produce ill-formed inputs that fail to reflect realistic data distributions, struggle to capture context-sensitive probabilistic dependencies, and lack explainability. We introduce ExplainFuzz, a test generation framework that leverages Probabilistic Circuits (PCs) to learn and query structured distributions over grammar-based test inputs interpretably and controllably. Starting from a Context-Free Grammar (CFG), ExplainFuzz compiles a grammar-aware PC and trains it on existing inputs. New inputs are then generated via sampling. ExplainFuzz utilizes the conditioning capability of PCs to incorporate test-specific constraints (e.g., a query must have GROUP BY), enabling constrained probabilistic sampling to generate inputs satisfying grammar and user-provided constraints. Our results show that ExplainFuzz improves the coherence and realism of generated inputs, achieving significant perplexity reduction compared to pCFGs, grammar-unaware PCs, and LLMs. By leveraging its native conditioning capability, ExplainFuzz significantly enhances the diversity of inputs that satisfy a user-provided constraint. Compared to grammar-aware mutational fuzzing, ExplainFuzz increases bug-triggering rates from 35% to 63% in SQL and from 10% to 100% in XML. These results demonstrate the power of a learned input distribution over mutational fuzzing, which is often limited to exploring the local neighborhood of seed inputs. These capabilities highlight the potential of PCs to serve as a foundation for grammar-aware, controllable test generation that captures context-sensitive, probabilistic dependencies.

We present the first systematic study of when target context helps molecular property prediction, evaluating context conditioning across 10 diverse protein families, 4 fusion architectures, data regimes spanning 67-9,409 training compounds, and both temporal and random evaluation splits. Using NestDrug, a FiLM-based architecture that conditions molecular representations on target identity, we characterize both success and failure modes with three principal findings. First, fusion architecture dominates: FiLM outperforms concatenation by 24.2 percentage points and additive conditioning by 8.6 pp; how you incorporate context matters more than whether you include it. Second, context enables otherwise impossible predictions: on data-scarce CYP3A4 (67 training compounds), multi-task transfer achieves 0.686 AUC where per-target Random Forest collapses to 0.238. Third, context can systematically hurt: distribution mismatch causes 10.2 pp degradation on BACE1; few-shot adaptation consistently underperforms zero-shot. Beyond methodology, we expose fundamental flaws in standard benchmarking: 1-nearest-neighbor Tanimoto achieves 0.991 AUC on DUD-E without any learning, and 50% of actives leak from training data, rendering absolute performance metrics meaningless. Our temporal split evaluation (train up to 2020, test 2021-2024) achieves stable 0.843 AUC with no degradation, providing the first rigorous evidence that context-conditional molecular representations generalize to future chemical space.

Misinformation is on the rise, and the strong writing capabilities of LLMs lower the barrier for malicious actors to produce and disseminate false information. We study how LLMs behave when prompted to spread misinformation across languages and target countries, and introduce GlobalLies, a multilingual parallel dataset of 440 misinformation generation prompt templates and 6,867 entities, spanning 8 languages and 195 countries. Using both human annotations and large-scale LLM-as-a-judge evaluations across hundreds of thousands of generations from state-of-the-art models, we show that misinformation generation varies systematically based on the country being discussed. Propagation of lies by LLMs is substantially higher in many lower-resource languages and for countries with a lower Human Development Index (HDI). We find that existing mitigation strategies provide uneven protection: input safety classifiers exhibit cross-lingual gaps, and retrieval-augmented fact-checking remains inconsistent across regions due to unequal information availability. We release GlobalLies for research purposes, aiming to support the development of mitigation strategies to reduce the spread of global misinformation: https://github.com/zohaib-khan5040/globallies

Large language models show potential for scalable mental-health support by simulating Cognitive Behavioral Therapy (CBT) counselors. However, existing methods often rely on static cognitive profiles and omniscient single-agent simulation, failing to capture the dynamic, information-asymmetric nature of real therapy. We introduce CCD-CBT, a multi-agent framework that shifts CBT simulation along two axes: 1) from a static to a dynamically reconstructed Cognitive Conceptualization Diagram (CCD), updated by a dedicated Control Agent, and 2) from omniscient to information-asymmetric interaction, where the Therapist Agent must reason from inferred client states. We release CCDCHAT, a synthetic multi-turn CBT dataset generated under this framework. Evaluations with clinical scales and expert therapists show that models fine-tuned on CCDCHAT outperform strong baselines in both counseling fidelity and positive-affect enhancement, with ablations confirming the necessity of dynamic CCD guidance and asymmetric agent design. Our work offers a new paradigm for building theory-grounded, clinically-plausible conversational agents.

OpenClaw's ClawHub marketplace hosts over 13,000 community-contributed agent skills, and between 13% and 26% of them contain security vulnerabilities according to recent audits. Regex scanners miss obfuscated payloads; formal static analyzers cannot read the natural language instructions in SKILL.md files where prompt injection and social engineering attacks hide. Neither approach handles both modalities. SkillSieve is a three-layer detection framework that applies progressively deeper analysis only where needed. Layer 1 runs regex, AST, and metadata checks through an XGBoost-based feature scorer, filtering roughly 86% of benign skills in under 40ms on average at zero API cost. Layer 2 sends suspicious skills to an LLM, but instead of asking one broad question, it splits the analysis into four parallel sub-tasks (intent alignment, permission justification, covert behavior detection, cross-file consistency), each with its own prompt and structured output. Layer 3 puts high-risk skills before a jury of three different LLMs that vote independently and, if they disagree, debate before reaching a verdict. We evaluate on 49,592 real ClawHub skills and adversarial samples across five evasion techniques, running the full pipeline on a 440 ARM single-board computer. On a 400-skill labeled benchmark, SkillSieve achieves 0.800 F1, outperforming ClawVet's 0.421, at an average cost of 0.006 per skill. Code, data, and benchmark are open-sourced.

In this work, we demonstrate that reliable stochastic sampling is a fundamental yet unfulfilled requirement for Large Language Models (LLMs) operating as agents. Agentic systems are frequently required to sample from distributions, often inferred from observed data, a process which needs to be emulated by the LLM. This leads to a distinct failure point: while standard RL agents rely on external sampling mechanisms, LLMs fail to map their internal probability estimates to their stochastic outputs. Through rigorous empirical analysis across multiple model families, model sizes, prompting styles, and distributions, we demonstrate the extent of this failure. Crucially, we show that while powerful frontier models can convert provided random seeds to target distributions, their ability to sample directly from specific distributions is fundamentally flawed.

Empirical scaling laws for language models have encouraged the development of ever-larger LLMs, despite their growing computational and memory costs. Sparse Mixture-of-Experts (MoEs) offer a promising alternative by activating only a subset of experts per forward pass, improving efficiency without sacrificing performance. However, the large number of expert parameters still leads to substantial memory consumption. Existing pruning methods typically allocate budgets uniformly across layers, overlooking the heterogeneous redundancy that arises in sparse MoEs. We propose GRAPE (Global Redundancy-Aware Pruning of Experts, a global pruning strategy that dynamically allocates pruning budgets based on cross-layer redundancy. Experiments on Mixtral-8x7B, Mixtral-8x22B, DeepSeek-MoE, Qwen-MoE, and GPT-OSS show that, under the same pruning budget, GRAPE consistently achieves the best average performance. On the three main models reported in the paper, it improves average accuracy over the strongest local baseline by 1.40% on average across pruning settings, with gains of up to 2.45%.

We investigate whether a multiscale tensor-network architecture can provide a useful inductive bias for reconstruction-based anomaly detection in collider jets. Jets are produced by a branching cascade, so their internal structure is naturally organised across angular and momentum scales. This motivates an autoencoder that compresses information hierarchically and can reorganise short-range correlations before coarse-graining. Guided by this picture, we formulate a MERA-inspired autoencoder acting directly on ordered jet constituents. To the best of our knowledge, a MERA-inspired autoencoder has not previously been proposed, and this architecture has not been explored in collider anomaly detection. We compare this architecture to a dense autoencoder, the corresponding tree-tensor-network limit, and standard classical baselines within a common background-only reconstruction framework. The paper is organised around two main questions: whether locality-aware hierarchical compression is genuinely supported by the data, and whether the disentangling layers of MERA contribute beyond a simpler tree hierarchy. To address these questions, we combine benchmark comparisons with a training-free local-compressibility diagnostic and a direct identity-disentangler ablation. The resulting picture is that the locality-preserving multiscale structure is well matched to jet data, and that the MERA disentanglers become beneficial precisely when the compression bottleneck is strongest. Overall, the study supports locality-aware hierarchical compression as a useful inductive bias for jet anomaly detection.

In this paper, we propose a novel framework for non-stationary time-series analysis that replaces conventional correlation-based statistics with direct estimation of statistical dependence in the normalized joint density of input and target signals, the cross density ratio (CDR). Unlike windowed correlation estimates, this measure is independent of sample order and robust to regime changes. The method builds on the functional maximal correlation algorithm (FMCA), which constructs a projection space by decomposing the eigenspectrum of the CDR. Multiscale features from this eigenspace are classified using a lightweight single-hidden-layer perceptron. On the TI-46 digit speech corpus, our approach outperforms hidden Markov models (HMMs) and state-of-the-art spiking neural networks, achieving higher accuracy with fewer than 10 layers and a storage footprint under 5 MB.

The mean-field Schrödinger bridge (MFSB) problem concerns designing a minimum-effort controller that guides a diffusion process with nonlocal interaction to reach a given distribution from another by a fixed deadline. Unlike the standard Schrödinger bridge, the dynamical constraint for MFSB is the mean-field limit of a population of interacting agents with controls. It serves as a natural model for large-scale multi-agent systems. The MFSB is computationally challenging because the nonlocal interaction makes the problem nonconvex. We propose a generalization of the Hopf-Cole transform for MFSB and, building on it, design a Sinkhorn-type recursive algorithm to solve the associated system of integro-PDEs. Under mild assumptions on the interaction potential, we discuss convergence guarantees for the proposed algorithm. We present numerical examples with repulsive and attractive interactions to illustrate the theoretical contributions.

Achieving optimal rates for stochastic composite convex optimization without prior knowledge of problem parameters remains a central challenge. In the deterministic setting, the auto-conditioned fast gradient method has recently been proposed to attain optimal accelerated rates without line-search procedures or prior knowledge of the Lipschitz smoothness constant, providing a natural prototype for parameter-free acceleration. However, extending this approach to the stochastic setting has proven technically challenging and remains open. Existing parameter-free stochastic methods either fail to achieve accelerated rates or rely on restrictive assumptions, such as bounded domains, bounded gradients, prior knowledge of the iteration horizon, or strictly sub-Gaussian noise. To address these limitations, we propose a stochastic variant of the auto-conditioned fast gradient method, referred to as stochastic AC-FGM. The proposed method is fully adaptive to the Lipschitz constant, the iteration horizon, and the noise level, enabling both adaptive stepsize selection and adaptive mini-batch sizing without line-search procedures. Under standard bounded conditional variance assumptions, we show that stochastic AC-FGM achieves the optimal iteration complexity of $O(1/\sqrt{\varepsilon})$ and the optimal sample complexity of $O(1/\varepsilon^2)$.

Quantum operations on pure states can be fully represented by unitary matrices. Variational quantum circuits, also known as quantum neural networks, embed data and trainable parameters into gate-based operations and optimize the parameters via gradient descent. The high cost of training and low fidelity of current quantum devices, however, restricts much of quantum machine learning to classical simulation. For few-qubit problems with large datasets, training the matrix elements directly, as is done with weight matrices in classical neural networks, can be faster than decomposing data and parameters into gates. We propose a method that trains matrices directly while maintaining unitarity through a single regularization term added to the loss function. A second training step, circuit alignment, then recovers a gate-based architecture from the resulting soft-unitary. On a five-qubit supervised classification task with 1000 datapoints, this two-step process produces a trained variational circuit in under four minutes, compared to over two hours for direct circuit training, while achieving lower binary cross-entropy loss. In a second experiment, soft-unitaries are embedded in a hybrid quantum-classical network for a reinforcement learning cartpole task, where the hybrid agent outperforms a purely classical baseline of comparable size.

We address the problems of giving a semantics to- and doing query answering (QA) on a relational database (RDB) that has missing values (MVs). The causes for the latter are governed by a Missingness Mechanism that is modelled as a Bayesian Network, which represents a Missingness Graph (MG) and involves the DB attributes. Our approach considerable departs from the treatment of RDBs with NULL (values). The MG together with the observed DB allow to build a block-independent probabilistic DB, on which basis we propose two QA techniques that jointly capture probabilistic uncertainty and statistical plausibility of the implicit imputation of MVs. We obtain complexity results that characterize the computational feasibility of those approaches.

Large volumes of medical data remain underutilized because centralizing distributed data is often infeasible due to strict privacy regulations and institutional constraints. In addition, models trained in centralized settings frequently fail to generalize across clinical sites because of heterogeneity in imaging protocols and continuously evolving data distributions arising from differences in scanners, acquisition parameters, and patient populations. Federated learning offers a promising solution by enabling collaborative model training without sharing raw data. However, incorporating differential privacy into federated learning, while essential for privacy guarantees, often leads to degraded accuracy, unstable convergence, and reduced generalization. In this work, we propose an adaptive differentially private federated learning (ADP-FL) framework for medical image segmentation that dynamically adjusts privacy mechanisms to better balance the privacy-utility trade-off. The proposed approach stabilizes training, significantly improves Dice scores and segmentation boundary quality, and maintains rigorous privacy guarantees. We evaluated ADP-FL across diverse imaging modalities and segmentation tasks, including skin lesion segmentation in dermoscopic images, kidney tumor segmentation in 3D CT scans, and brain tumor segmentation in multi-parametric MRI. Compared with conventional federated learning and standard differentially private federated learning, ADP-FL consistently achieves higher accuracy, improved boundary delineation, faster convergence, and greater training stability, with performance approaching that of non-private federated learning under the same privacy budgets. These results demonstrate the practical viability of ADP-FL for high-performance, privacy-preserving medical image segmentation in real-world federated settings.

Sparse Mixture-of-Experts (MoE) allows scaling of language and vision models efficiently by activating only a small subset of experts per input. While this reduces computation, the large number of parameters still incurs substantial memory overhead during inference. Post-training quantization has been explored to address this issue. Because uniform quantization suffers from significant accuracy loss at low bit-widths, mixed-precision methods have been recently explored; however, they often require substantial computation for bit-width allocation and overlook the varying sensitivity of model performance to the quantization of different experts. We propose a theoretically grounded expert-wise mixed precision strategy that assigns bit-width to each expert primarily based on their change in routers l2 norm during training. Experts with smaller changes are shown to capture less frequent but critical features, and model performance is more sensitive to the quantization of these experts, thus requiring higher precision. Furthermore, to avoid allocating experts to lower precision that inject high quantization noise, experts with large maximum intra-neuron variance are also allocated higher precision. Experiments on large-scale MoE models, including Switch Transformer and Mixtral, show that our method achieves higher accuracy than existing approaches, while also reducing inference cost and incurring only negligible overhead for bit-width assignment.

Pashto is absent from Whisper's pre-training corpus despite being one of CommonVoice's largest language collections, leaving off-the-shelf models unusable: all Whisper sizes output Arabic, Dari, or Urdu script on Pashto audio, achieving word error rates above 100%. We compare four fine-tuning strategies for whisper-base on CommonVoice Pashto v20: vanilla full fine-tuning, LoRA (rank 64), frozen-encoder (2/6 layers), and multistage Urdu-to-Pashto transfer. We extend vanilla fine-tuning to whisper-small and whisper-large-v3-turbo on CommonVoice Pashto v24 (113 hours). Vanilla fine-tuning achieves WER 21.22% on CV20, outperforming LoRA by 33.36 pp, frozen-encoder by 14.76 pp, and Urdu transfer by 44.56 pp. Frozen-encoder fine-tuning degrades performance on whisper-base (6 encoder layers): layer-function separation does not hold at this depth, and freezing removes a third of trainable capacity. Urdu-to-Pashto transfer fails due to an unverified intermediate checkpoint, phonological mismatch, and insufficient training. On CV24, whisper-small achieves WER 24.89% (2.24 pp over whisper-base at 3.3x parameters); whisper-large-v3-turbo achieves 23.37% (a further 1.52 pp). Diminishing returns indicate whisper-small is the practical optimum at 113 hours. Online augmentation provides 7.25 pp WER benefit over matched training. Error analysis identifies word-final suffix confusion (masculine -ay vs. feminine -a) and retroflex substitutions involving the Pashto-unique consonant /ts/ as dominant failure modes. Fine-tuned checkpoints and evaluation scripts are released on HuggingFace.

Symbolic execution detects vulnerabilities with precision, but applying it to large codebases requires harnesses that set up symbolic state, model dependencies, and specify assertions. Writing these harnesses has traditionally been a manual process requiring expert knowledge, which significantly limits the scalability of the technique. We present Static Analysis Informed and LLM-Orchestrated Symbolic Execution (SAILOR), which automates symbolic execution harness construction by combining static analysis with LLM-based synthesis. SAILOR operates in three phases: (1) static analysis identifies candidate vulnerable locations and generates vulnerability specifications; (2) an LLM uses vulnerability specifications and orchestrates harness synthesis by iteratively refining drivers, stubs, and assertions against compiler and symbolic execution feedback; symbolic execution then detects vulnerabilities using the generated harness, and (3) concrete replay validates the symbolic execution results against the unmodified project source. This design combines the scalability of static analysis, the code reasoning of LLMs, the path precision of symbolic execution, and the ground truth produced by concrete execution. We evaluate SAILOR on 10 open-source C/C++ projects totaling 6.8 M lines of code. SAILOR discovers 379 distinct, previously unknown memory-safety vulnerabilities (421 confirmed crashes). The strongest of five baselines we compare SAILOR to (agentic vulnerability detection using Claude Code with full codebase access and unlimited interaction), finds only 12 vulnerabilities. Each phase of SAILOR is critical: Without static analysis targeting confirmed vulnerabilities drop 12.2X; without iterative LLM synthesis zero vulnerabilities are confirmed; and without symbolic execution no approach can detect more than 12 vulnerabilities.

Large language models (LLMs) are widely explored for reasoning-intensive research tasks, yet resources for testing whether they can infer scientific conclusions from structured biomedical evidence remain limited. We introduce $\textbf{MedConclusion}$, a large-scale dataset of $\textbf{5.7M}$ PubMed structured abstracts for biomedical conclusion generation. Each instance pairs the non-conclusion sections of an abstract with the original author-written conclusion, providing naturally occurring supervision for evidence-to-conclusion reasoning. MedConclusion also includes journal-level metadata such as biomedical category and SJR, enabling subgroup analysis across biomedical domains. As an initial study, we evaluate diverse LLMs under conclusion and summary prompting settings and score outputs with both reference-based metrics and LLM-as-a-judge. We find that conclusion writing is behaviorally distinct from summary writing, strong models remain closely clustered under current automatic metrics, and judge identity can substantially shift absolute scores. MedConclusion provides a reusable data resource for studying scientific evidence-to-conclusion reasoning. Our code and data are available at: https://github.com/Harvard-AI-and-Robotics-Lab/MedConclusion.

Vision-Language Models (VLMs) face significant safety vulnerabilities from malicious prompt attacks due to weakened alignment during visual integration. Existing defenses suffer from efficiency and robustness. To address these challenges, we first propose the Multimodal Aggregated Feature Extraction (MAFE) framework that enables CLIP to handle long text and fuse multimodal information into unified representations. Through empirical analysis of MAFE-extracted features, we discover distinct distributional patterns between benign and malicious prompts. Building upon this finding, we develop VLMShield, a lightweight safety detector that efficiently identifies multimodal malicious attacks as a plug-and-play solution. Extensive experiments demonstrate superior performance across multiple dimensions, including robustness, efficiency, and utility. Through our work, we hope to pave the way for more secure multimodal AI deployment. Code is available at [this https URL](https://github.com/pgqihere/VLMShield).

Analogical reasoning is a hallmark of human intelligence, enabling us to solve new problems by transferring knowledge from one situation to another. Yet, developing artificial intelligence systems capable of robust human-like analogical reasoning has proven difficult. In this work, we train transformers using Meta-Learning for Compositionality (MLC) on an analogical reasoning task (letter-string analogies) and assess their generalization capabilities. We find that letter-string analogies become learnable when guiding the models to attend to the most informative problem elements induced by including copying tasks in the training data. Furthermore, generalization to new alphabets becomes better when models are trained with more heterogeneous datasets, where our 3-layer encoder-decoder model outperforms most frontier models. The MLC approach also enables some generalization to compositions of trained transformations, but not to completely novel transformations. To understand how the model operates, we identify an algorithm that approximates the model's computations. We verify this using interpretability analyses and show that the model can be steered precisely according to expectations derived from the algorithm. Finally, we discuss implications of our findings for generalization capabilities of larger models and parallels to human analogical reasoning.

Sparse autoencoders (SAEs) are widely used in mechanistic interpretability to project LLM activations onto sparse latent spaces. However, sparsity alone is an imperfect proxy for interpretability, and current training objectives often result in brittle latent representations. SAEs are known to be prone to feature absorption, where general features are subsumed by more specific ones due to co-occurrence, degrading interpretability despite high reconstruction fidelity. Recent negative results on Out-of-Distribution (OOD) performance further underscore broader robustness related failures tied to under-specified training objectives. We address this by proposing a masking-based regularization that randomly replaces tokens during training to disrupt co-occurrence patterns. This improves robustness across SAE architectures and sparsity levels reducing absorption, enhancing probing performance, and narrowing the OOD gap. Our results point toward a practical path for more reliable interpretability tools.

We study stochastic convex optimization (SCO) with heavy-tailed gradients under pure epsilon-differential privacy (DP). Instead of assuming a bound on the worst-case Lipschitz parameter of the loss, we assume only a bounded k-th moment. This assumption allows for unbounded, heavy-tailed stochastic gradient distributions, and can yield sharper excess risk bounds. The minimax optimal rate for approximate (epsilon, delta)-DP SCO is known in this setting, but the pure epsilon-DP case has remained open. We characterize the minimax optimal excess-risk rate for pure epsilon-DP heavy-tailed SCO up to logarithmic factors. Our algorithm achieves this rate in polynomial time with high probability. Moreover, it runs in polynomial time with probability 1 when the worst-case Lipschitz parameter is polynomially bounded. For important structured problem classes - including hinge/ReLU-type and absolute-value losses on Euclidean balls, ellipsoids, and polytopes - we achieve the same excess-risk guarantee in polynomial time with probability 1 even when the worst-case Lipschitz parameter is infinite. Our approach is based on a novel framework for privately optimizing Lipschitz extensions of the empirical loss. We complement our excess risk upper bound with a novel high probability lower bound.

We introduce Discrete flow Matching policy Optimization (DoMinO), a unified framework for Reinforcement Learning (RL) fine-tuning Discrete Flow Matching (DFM) models under a broad class of policy gradient methods. Our key idea is to view the DFM sampling procedure as a multi-step Markov Decision Process. This perspective provides a simple and transparent reformulation of fine-tuning reward maximization as a robust RL objective. Consequently, it not only preserves the original DFM samplers but also avoids biased auxiliary estimators and likelihood surrogates used by many prior RL fine-tuning methods. To prevent policy collapse, we also introduce new total-variation regularizers to keep the fine-tuned distribution close to the pretrained one. Theoretically, we establish an upper bound on the discretization error of DoMinO and tractable upper bounds for the regularizers. Experimentally, we evaluate DoMinO on regulatory DNA sequence design. DoMinO achieves stronger predicted enhancer activity and better sequence naturalness than the previous best reward-driven baselines. The regularization further improves alignment with the natural sequence distribution while preserving strong functional performance. These results establish DoMinO as an useful framework for controllable discrete sequence generation.

Conventional end-to-end automatic speech recognition (ASR) systems rely on paired speech-text data for domain adaptation. Recent LLM-based ASR architectures connect a speech encoder to a large language model via a projection module, enabling adaptation with text-only data. However, this introduces a modality gap, as the LLM is not exposed to the noisy representations produced by the speech projector. We investigate whether small amounts of speech can mitigate this mismatch. We compare three strategies: text-only adaptation, paired speech-text adaptation, and mixed batching (MB), which combines both. Experiments in in-domain and out-of-domain settings show that even limited speech consistently improves performance. Notably, MB using only 10% of the target-domain (less than 4 hours) speech achieves word error rates comparable to, or better than, conventional ASR fine-tuning with the full dataset, indicating that small amounts of speech provide a strong modality-alignment signal.

"Best-of-N" selection is a popular inference-time scaling method for code generation using Large Language Models (LLMs). However, to reliably identify correct solutions, existing methods often depend on expensive or stochastic external verifiers. In this paper, we propose Symbolic Equivalence Partitioning, a selection framework that uses symbolic execution to group candidate programs by semantic behavior and select a representative from the dominant functional partition. To improve grouping and selection, we encode domain-specific constraints as Satisfiability Modulo Theories (SMT) assumptions during symbolic execution to reduce path explosion and prevent invalid input searches outside the problem domain. At N=10, our method improves average accuracy over Pass@1 from 0.728 to 0.803 on HumanEval+ and from 0.516 to 0.604 on LiveCodeBench, without requiring any additional LLM inference beyond the initial N candidate generations.

Cultural values are expressed not only through language but also through visual scenes and everyday social practices. Yet existing evaluations of cultural values in language models are almost entirely text-only, making it unclear whether models can ground culture-conditioned judgments when response options are visualized. We introduce ValueGround, a benchmark for evaluating culture-conditioned visual value grounding in multimodal large language models (MLLMs). Built from World Values Survey (WVS) questions, ValueGround uses minimally contrastive image pairs to represent opposing response options while controlling irrelevant variation. Given a country, a question, and an image pair, a model must choose the image that best matches the country's value tendency without access to the original response-option texts. Across six MLLMs and 13 countries, average accuracy drops from 72.8% in the text-only setting to 65.8% when options are visualized, despite 92.8% accuracy on option-image alignment. Stronger models are more robust, but all remain prone to prediction reversals. Our benchmark provides a controlled testbed for studying cross-modal transfer of culture-conditioned value judgments.

Large language models that require multiple GPU cards to host are usually the most capable models. It is necessary to understand and steer these models, but the current technologies do not support the interpretability and steering of these models in the multi-GPU setting as well as the single-GPU setting. We present a practical implementation of activation-level interpretability (logit lens) and steering (steering vector) that scales up to multi-GPU language models. Our system implements design choices that reduce the activation memory by up to 7x and increase the throughput by up to 41x compared to a baseline on identical hardware. We demonstrate the method across LLaMA-3.1 (8B, 70B) and Qwen-3 (4B, 14B, 32B), sustaining 20-100 tokens/s while collecting full layer-wise activation trajectories for sequences of 1,500 tokens. Using label-position steering vectors injected post-LayerNorm, we show controllable, monotonic shifts in model outputs with a mean steerability slope of 0.702 across evaluated datasets, without fine-tuning or additional forward passes. We release detailed benchmarks, ablations, and a reproducible instrumentation recipe to enable practical interpretability and real-time behavioral control for frontier LLMs at https://github.com/Devdesai1901/LogitLense.

Time-resolved volumetric MR imaging that reconstructs a 3D MRI within sub-seconds to resolve deformable motion is essential for motion-adaptive radiotherapy. Representing patient anatomy and associated motion fields as 3D Gaussians, we developed a spatiotemporal Gaussian representation-based framework (DREME-GSMR), which enables time-resolved dynamic MRI reconstruction from a pre-treatment 3D MR scan without any prior anatomical/motion model. DREME-GSMR represents a reference MRI volume and a corresponding low-rank motion model (as motion-basis components) using 3D Gaussians, and incorporates a dual-path MLP/CNN motion encoder to estimate temporal motion coefficients of the motion model from raw k-space-derived signals. Furthermore, using the solved motion model, DREME-GSMR can infer motion coefficients directly from new online k-space data, allowing subsequent intra-treatment volumetric MR imaging and motion tracking (real-time imaging). A motion-augmentation strategy is further introduced to improve robustness to unseen motion patterns during real-time imaging. DREME-GSMR was evaluated on the XCAT digital phantom, a physical motion phantom, and MR-LINAC datasets acquired from 6 healthy volunteers and 20 patients (with independent sequential scans for cross-evaluation). DREME-GSMR reconstructs MRIs of a ~400ms temporal resolution, with an inference time of ~10ms/volume. In XCAT experiments, DREME-GSMR achieved mean(s.d.) SSIM, tumor center-of-mass-error(COME), and DSC of 0.92(0.01)/0.91(0.02), 0.50(0.15)/0.65(0.19) mm, and 0.92(0.02)/0.92(0.03) for dynamic reconstruction/real-time imaging. For the physical phantom, the mean target COME was 1.19(0.94)/1.40(1.15) mm for dynamic/real-time imaging, while for volunteers and patients, the mean liver COME for real-time imaging was 1.31(0.82) and 0.96(0.64) mm, respectively.

This study introduces a hybrid deep learning model for intrusion detection in Industrial IoT (IIoT) systems, combining ResNet-1D, BiGRU, and Multi-Head Attention (MHA) for effective spatial-temporal feature extraction and attention-based feature weighting. To address class imbalance, SMOTE was applied during training on the EdgeHoTset dataset. The model achieved 98.71% accuracy, a loss of 0.0417%, and low inference latency (0.0001 sec /instance), demonstrating strong real-time capability. To assess generalizability, the model was also tested on the CICIoV2024 dataset, where it reached 99.99% accuracy and F1-score, with a loss of 0.0028, 0 % FPR, and 0.00014 sec/instance inference time. Across all metrics and datasets, the proposed model outperformed existing methods, confirming its robustness and effectiveness for real-time IoT intrusion detection.

Deep Learning Reduced Order Models (ROMs) are becoming increasingly popular as surrogate models for parametric partial differential equations (PDEs) due to their ability to handle high-dimensional data, approximate highly nonlinear mappings, and utilize GPUs. Existing approaches typically learn evolution either on the full solution field, which requires capturing long-range spatial interactions at high computational cost, or on compressed latent representations obtained from autoencoders, which reduces the cost but often yields latent vectors that are difficult to evolve, since they primarily encode spatial information. Moreover, in parametric PDEs, the initial condition alone is not sufficient to determine the trajectory, and most current approaches are not evaluated on jointly predicting multiple solution components with differing magnitudes and parameter sensitivities. To address these challenges, we propose a joint model consisting of a convolutional encoder, a transformer operating on latent representations, and a decoder for reconstruction. The main novelties are joint training with multi-stage parameter injection and coordinate channel injection. Parameters are injected at multiple stages to improve conditioning. Physical coordinates are encoded to provide spatial information. This allows the model to dynamically adapt its computations to the specific PDE parameters governing each system, rather than learning a single fixed response. Experiments on the Advection-Diffusion-Reaction equation and Navier-Stokes flow around the cylinder wake demonstrate that our approach combines the efficiency of latent evolution with the fidelity of full-field models, outperforming DL-ROMs, latent transformers, and plain ViTs in multi-field prediction, reducing the relative rollout error by approximately $5$ times.

Deep research with Large Language Model (LLM) agents is emerging as a powerful paradigm for multi-step information discovery, synthesis, and analysis. However, existing approaches primarily focus on unstructured web data, while the challenges of conducting deep research over large-scale structured databases remain relatively underexplored. Unlike web-based research, effective data-centric research requires more than retrieval and summarization and demands iterative hypothesis generation, quantitative reasoning over structured schemas, and convergence toward a coherent analytical narrative. In this paper, we present DataSTORM, an LLM-based agentic system capable of autonomously conducting research across both large-scale structured databases and internet sources. Grounded in principles from Exploratory Data Analysis and Data Storytelling, DataSTORM reframes deep research over structured data as a thesis-driven analytical process: discovering candidate theses from data, validating them through iterative cross-source investigation, and developing them into coherent analytical narratives. We evaluate DataSTORM on InsightBench, where it achieves a new state-of-the-art result with a 19.4% relative improvement in insight-level recall and 7.2% in summary-level score. We further introduce a new dataset built on ACLED, a real-world complex database, and demonstrate that DataSTORM outperforms proprietary systems such as ChatGPT Deep Research across both automated metrics and human evaluations.

Multivariate forecasting with Transformers faces a core scalability challenge: modeling cross-channel dependencies via attention compounds attention's quadratic sequence complexity with quadratic channel scaling, making full cross-channel attention impractical for high-dimensional time series. We propose Multivariate Infini Compressive Attention (MICA), an architectural design to extend channel-independent Transformers to channel-dependent forecasting. By adapting efficient attention techniques from the sequence dimension to the channel dimension, MICA adds a cross-channel attention mechanism to channel-independent backbones that scales linearly with channel count and context length. We evaluate channel-independent Transformer architectures with and without MICA across multiple forecasting benchmarks. MICA reduces forecast error over its channel-independent counterparts by 5.4% on average and up to 25.4% on individual datasets, highlighting the importance of explicit cross-channel modeling. Moreover, models with MICA rank first among deep multivariate Transformer and MLP baselines. MICA models also scale more efficiently with respect to both channel count and context length than Transformer baselines that compute attention across both the temporal and channel dimensions, establishing compressive attention as a practical solution for scalable multivariate forecasting.

Most methods for learning with noisy labels require privileged knowledge such as noise transition matrices, clean subsets or pretrained feature extractors, resources typically unavailable when robustness is most needed. We propose Conformal Margin Risk Minimization (CMRM), a plug-and-play envelope framework that improves any classification loss under label noise by adding a single quantile-calibrated regularization term, with no privileged knowledge or training pipeline modification. CMRM measures the confidence margin between the observed label and competing labels, and thresholds it with a conformal quantile estimated per batch to focus training on high-margin samples while suppressing likely mislabeled ones. We derive a learning bound for CMRM under arbitrary label noise requiring only mild regularity of the margin distribution. Across five base methods and six benchmarks with synthetic and real-world noise, CMRM consistently improves accuracy (up to +3.39%), reduces conformal prediction set size (up to -20.44%) and does not hurt under 0% noise, showing that CMRM captures a method-agnostic uncertainty signal that existing mechanisms did not exploit.

Reasoning models have demonstrated remarkable capabilities in solving complex problems by leveraging long chains of thought. However, this more deliberate reasoning comes with substantial computational overhead at inference time. The Long-to-Short (L2S) reasoning problem seeks to maintain high accuracy using fewer tokens, but current training-free model merging approaches rely on scalarized, fixed-hyperparameter arithmetic methods that are highly brittle and force suboptimal compromises. To address this gap, we introduce Evo-L2S, a novel framework that formulates L2S reasoning as a multi-objective optimization challenge. By leveraging evolutionary model merging, Evo-L2S explicitly optimizes the trade-off between accuracy and output length to produce a robust Pareto front of merged models. To make this search computationally tractable for large language models, we propose an entropy-based subset sampling technique that drastically reduces the overhead of fitness estimation. Comprehensive experiments across 1.5B, 7B, and 14B parameter scales on six mathematical reasoning benchmarks demonstrate that Evo-L2S can reduce the length of generated reasoning traces by over 50% while preserving, or even improving, the problem-solving accuracy of the original reasoning models.

Conformal prediction provides distribution-free prediction intervals with finite-sample coverage guarantees, and recent work by Snell \& Griffiths reframes it as Bayesian Quadrature (BQ-CP), yielding powerful data-conditional guarantees via Dirichlet posteriors over thresholds. However, BQ-CP fundamentally requires the i.i.d. assumption -- a limitation the authors themselves identify. Meanwhile, weighted conformal prediction handles distribution shift via importance weights but remains frequentist, producing only point-estimate thresholds. We propose \textbf{Weighted Bayesian Conformal Prediction (WBCP)}, which generalizes BQ-CP to arbitrary importance-weighted settings by replacing the uniform Dirichlet $\Dir(1,\ldots,1)$ with a weighted Dirichlet $\Dir(\neff \cdot \tilde{w}_1, \ldots, \neff \cdot \tilde{w}_n)$, where $\neff$ is Kish's effective sample size. We prove four theoretical results: (1)~$\neff$ is the unique concentration parameter matching frequentist and Bayesian variances; (2)~posterior standard deviation decays as $O(1/\sqrt{\neff})$; (3)~BQ-CP's stochastic dominance guarantee extends to per-weight-profile data-conditional guarantees; (4)~the HPD threshold provides $O(1/\sqrt{\neff})$ improvement in conditional coverage. We instantiate WBCP for spatial prediction as \emph{Geographical BQ-CP}, where kernel-based spatial weights yield per-location posteriors with interpretable diagnostics. Experiments on synthetic and real-world spatial datasets demonstrate that WBCP maintains coverage guarantees while providing substantially richer uncertainty information.

Current Machine Translation (MT) systems for Arabic often struggle to account for dialectal diversity, frequently homogenizing dialectal inputs into Modern Standard Arabic (MSA) and offering limited user control over the target vernacular. In this work, we propose a context-aware and steerable framework for dialectal Arabic MT that explicitly models regional and sociolinguistic variation. Our primary technical contribution is a Rule-Based Data Augmentation (RBDA) pipeline that expands a 3,000-sentence seed corpus into a balanced 57,000-sentence parallel dataset, covering eight regional varieties eg., Egyptian, Levantine, Gulf, etc. By fine-tuning an mT5-base model conditioned on lightweight metadata tags, our approach enables controllable generation across dialects and social registers in the translation output. Through a combination of automatic evaluation and qualitative analysis, we observe an apparent accuracy-fidelity trade-off: high-resource baselines such as NLLB (No Language Left Behind) achieve higher aggregate BLEU scores (13.75) by defaulting toward the MSA mean, while exhibiting limited dialectal specificity. In contrast, our model achieves lower BLEU scores (8.19) but produces outputs that align more closely with the intended regional varieties. Supporting qualitative evaluation, including an LLM-assisted cultural authenticity analysis, suggests improved dialectal alignment compared to baseline systems (4.80/5 vs. 1.0/5). These findings highlight the limitations of standard MT metrics for dialect-sensitive tasks and motivate the need for evaluation practices that better reflect linguistic diversity in Arabic MT.

In nonlinear dynamical systems, tipping refers to a critical transition from one steady state to another, typically catastrophic, steady state, often resulting from a saddle-node bifurcation. Recently, the machine-learning framework of parameter-adaptable reservoir computing has been applied to predict tipping in systems described by low-dimensional stochastic differential equations. However, anticipating tipping in complex spatiotemporal dynamical systems remains a significant open problem. The ability to forecast not only the occurrence but also the precise timing of such tipping events is crucial for providing the actionable lead time necessary for timely mitigation. By utilizing the mathematical approach of non-negative matrix factorization to generate dimensionally reduced spatiotemporal data as input, we exploit parameter-adaptable reservoir computing to accurately anticipate tipping. We demonstrate that the tipping time can be identified within a narrow prediction window across a variety of spatiotemporal dynamical systems, as well as in CMIP5 (Coupled Model Intercomparison Project 5) climate projections. Furthermore, we show that this reservoir-computing framework, utilizing reduced input data, is robust against common forecasting challenges and significantly alleviates the computational overhead associated with processing full spatiotemporal data.

Multi-agent systems using large language models (LLMs) have demonstrated impressive capabilities across various domains. However, current agent communication suffers from verbose output that overload context and increase computational costs. Although existing approaches focus on compressing the message from the speaker side, they struggle to adapt to different listeners and identify relevant information. An effective way in human communication is to allow the listener to interrupt and express their opinion or ask for clarification. Motivated by this, we propose an interruptible communication framework that allows the agent who is listening to interrupt the current speaker. Through prompting experiments, we find that current LLMs are often overconfident and interrupt before receiving enough information. Therefore, we propose a learning method that predicts the appropriate interruption points based on the estimated future reward and cost. We evaluate our framework across various multi-agent scenarios, including 2-agent text pictionary games, 3-agent meeting scheduling, and 3-agent debate. The results of the experiment show that our HANDRAISER can reduce the communication cost by 32.2% compared to the baseline with comparable or superior task performance. This learned interruption behavior can also be generalized to different agents and tasks.

Manufacturing test flows in high-volume electronics production are typically fixed during product development and executed unchanged on every unit, even as failure patterns and process conditions evolve. This protects quality, but it also imposes unnecessary test cost, while existing data-driven methods mostly optimize static test subsets and neither adapt online to changing defect distributions nor explicitly control escape risk. In this study, we present an adaptive test-selection framework that combines offline minimum-cost diagnostic subset construction using greedy set cover with an online Thompson-sampling multi-armed bandit that switches between full and reduced test plans using a rolling process-stability signal. We evaluate the framework on two printed circuit board assembly stages-Functional Circuit Test and End-of-Line test-covering 28,000 board runs. Offline analysis identified zero-escape reduced plans that cut test time by 18.78% in Functional Circuit Test and 91.57\% in End-of-Line testing. Under temporal validation with real concept drift, static reduction produced 110 escaped defects in Functional Circuit Test and 8 in End-of-Line, whereas the adaptive policy reduced escapes to zero by reverting to fuller coverage when instability emerged in practice. These results show that online learning can preserve manufacturing quality while reducing test burden, offering a practical route to adaptive test planning across production domains, and offering both economic and logistics improvement for companies.

Prime Video regularly conducts load tests to simulate the viewer traffic spikes seen during live events such as Thursday Night Football as well as video-on-demand (VOD) events such as Rings of Power. While these stress tests validate system capacity, they can sometimes miss service behaviors unique to real event traffic. We present a graph-based anomaly detection system that identifies under-represented services using unsupervised node-level graph embeddings. Built on a GCN-GAE, our approach learns structural representations from directed, weighted service graphs at minute-level resolution and flags anomalies based on cosine similarity between load test and event embeddings. The system identifies incident-related services that are documented and demonstrates early detection capability. We also introduce a preliminary synthetic anomaly injection framework for controlled evaluation that show promising precision (96%) and low false positive rate (0.08%), though recall (58%) remains limited under conservative propagation assumptions. This framework demonstrates practical utility within Prime Video while also surfacing methodological lessons and directions, providing a foundation for broader application across microservice ecosystems.

Visual prompting (VP) has emerged as a popular method to repurpose pretrained vision models for adaptation to downstream tasks. Unlike conventional model fine-tuning techniques, VP introduces a universal perturbation directly into the input data to facilitate task-specific fine-tuning rather than modifying model parameters. However, there exists a noticeable performance gap between VP and conventional fine-tuning methods, highlighting an unexplored realm in theory and practice to understand and advance the input-level VP to reduce its current performance gap. Towards this end, we introduce a generalized concept, termed activation prompt (AP), which extends the scope of the input-level VP by enabling universal perturbations to be applied to activation maps within the intermediate layers of the model. By using AP to revisit the problem of VP and employing it as an analytical tool, we demonstrate the intrinsic limitations of VP in both performance and efficiency, revealing why input-level prompting may lack effectiveness compared to AP, which exhibits a model-dependent layer preference. We show that AP is closely related to normalization tuning in convolutional neural networks and vision transformers, although each model type has distinct layer preferences for prompting. We also theoretically elucidate the rationale behind such a preference by analyzing global features across layers. Through extensive experiments across 29 datasets and various model architectures, we provide a comprehensive performance analysis of AP, comparing it with VP and parameter-efficient fine-tuning baselines. Our results demonstrate AP's superiority in both accuracy and efficiency, considering factors such as time, parameters, memory usage, and throughput.

Forecasters often choose retraining schedules by convention rather than by an explicit decision rule. This paper gives that decision a posterior-space language. We define learning debt as the divergence between the deployed and continuously updated posteriors, define actionable staleness as the policy-relevant latent state, and derive a one-step Bayes retraining rule under an excess-loss formulation. In an online conjugate simulation using the exact Kullback-Leibler divergence between deployed and shadow normal-inverse-gamma posteriors, a debt-filter beats a default 10-period calendar baseline in 15 of 24 abrupt-shift cells, all 24 gradual-drift cells, and 17 of 24 variance-shift cells, and remains below the best fixed cadence in a grid of cadences (5, 10, 20, and 40 periods) in 10, 24, and 17 cells, respectively. Fixed-threshold CUSUM remains a strong benchmark, while a proxy filter built from indirect diagnostics performs poorly. A retrospective Airbnb production backtest shows how the same decision logic behaves around a known payment-policy shock.

We prove that no continuous, utility-preserving wrapper defense-a function $D: X\to X$ that preprocesses inputs before the model sees them-can make all outputs strictly safe for a language model with connected prompt space, and we characterize exactly where every such defense must fail. We establish three results under successively stronger hypotheses: boundary fixation-the defense must leave some threshold-level inputs unchanged; an $ε$-robust constraint-under Lipschitz regularity, a positive-measure band around fixed boundary points remains near-threshold; and a persistent unsafe region under a transversality condition, a positive-measure subset of inputs remains strictly unsafe. These constitute a defense trilemma: continuity, utility preservation, and completeness cannot coexist. We prove parallel discrete results requiring no topology, and extend to multi-turn interactions, stochastic defenses, and capacity-parity settings. The results do not preclude training-time alignment, architectural changes, or defenses that sacrifice utility. The full theory is mechanically verified in Lean 4 and validated empirically on three LLMs.

Visual Anomaly Detection (VAD) is a critical task for many applications including industrial inspection and healthcare. While VAD has been extensively studied, two key challenges remain largely unaddressed in conjunction: edge deployment, where computational resources are severely constrained, and continual learning, where models must adapt to evolving data distributions without forgetting previously acquired knowledge. Our benchmark provides guidance for the selection of the optimal backbone and VAD method under joint efficiency and adaptability constraints, characterizing the trade-offs between memory footprint, inference cost, and detection performance. Studying these challenges in isolation is insufficient, as methods designed for one setting make assumptions that break down when the other constraint is simultaneously imposed. In this work, we propose the first comprehensive benchmark for VAD on the edge in the continual learning scenario, evaluating seven VAD models across three lightweight backbone architectures. Furthermore, we propose Tiny-Dinomaly, a lightweight adaptation of the Dinomaly model built on the DINO foundation model that achieves 13x smaller memory footprint and 20x lower computational cost while improving Pixel F1 by 5 percentage points. Finally, we introduce targeted modifications to PatchCore and PaDiM to improve their efficiency in the continual learning setting.

Wave setup plays a significant role in transferring wave-induced energy to currents and causing an increase in water elevation. This excess momentum flux, known as radiation stress, motivates the coupling of circulation models with wave models to improve the accuracy of storm surge prediction, however, traditional numerical wave models are complex and computationally expensive. As a result, in practical coupled simulations, wave models are often executed at much coarser temporal resolution than circulation models. In this work, we explore the use of Deep Operator Networks (DeepONets) as a surrogate for the Simulating WAves Nearshore (SWAN) numerical wave model. The proposed surrogate model was tested on three distinct 1-D and 2-D steady-state numerical examples with variable boundary wave conditions and wind fields. When applied to a realistic numerical example of steady state wave simulation in Duck, NC, the model achieved consistently high accuracy in predicting the components of the radiation stress gradient and the significant wave height across representative scenarios.

The viability of chain-of-thought (CoT) monitoring hinges on models being unable to reason effectively in their latent representations. Yet little is known about the limits of such latent reasoning in LLMs. We test these limits by studying whether models can discover multi-step planning strategies without supervision on intermediate steps and execute them latently, within a single forward pass. Using graph path-finding tasks that precisely control the number of required latent planning steps, we uncover a striking limitation unresolved by massive scaling: tiny transformers trained from scratch discover strategies requiring up to three latent steps, fine-tuned GPT-4o and Qwen3-32B reach five, and GPT-5.4 attains seven under few-shot prompting. Although the maximum latent planning depth models can learn during training is five, the discovered strategy generalizes up to eight latent steps at test-time. This reveals a dissociation between the ability to discover a latent strategy under final-answer supervision alone and the ability to execute it once discovered. If similar limits hold more broadly, strategies requiring multiple coordinated latent planning steps may need to be explicitly taught or externalized, lending credence to CoT monitoring.

We propose a new frontier: Neural Computers (NCs) -- an emerging machine form that unifies computation, memory, and I/O in a learned runtime state. Unlike conventional computers, which execute explicit programs, agents, which act over external execution environments, and world models, which learn environment dynamics, NCs aim to make the model itself the running computer. Our long-term goal is the Completely Neural Computer (CNC): the mature, general-purpose realization of this emerging machine form, with stable execution, explicit reprogramming, and durable capability reuse. As an initial step, we study whether early NC primitives can be learned solely from collected I/O traces, without instrumented program state. Concretely, we instantiate NCs as video models that roll out screen frames from instructions, pixels, and user actions (when available) in CLI and GUI settings. These implementations show that learned runtimes can acquire early interface primitives, especially I/O alignment and short-horizon control, while routine reuse, controlled updates, and symbolic stability remain open. We outline a roadmap toward CNCs around these challenges. If overcome, CNCs could establish a new computing paradigm beyond today's agents, world models, and conventional computers.

This paper presents a transformer-based approach to solving the SympTEMIST named entity recognition (NER) and entity linking (EL) tasks. For NER, we fine-tune a RoBERTa-based (1) token-level classifier with BiLSTM and CRF layers on an augmented train set. Entity linking is performed by generating candidates using the cross-lingual SapBERT XLMR-Large (2), and calculating cosine similarity against a knowledge base. The choice of knowledge base proves to have the highest impact on model accuracy.

Understanding when Vision-Language Models (VLMs) will behave unexpectedly, whether models can reliably predict their own behavior, and if models adhere to their introspective reasoning are central challenges for trustworthy deployment. To study this, we introduce the Graded Color Attribution (GCA) dataset, a controlled benchmark designed to elicit decision rules and evaluate participant faithfulness to these rules. GCA consists of line drawings that vary pixel-level color coverage across three conditions: world-knowledge recolorings, counterfactual recolorings, and shapes with no color priors. Using GCA, both VLMs and human participants establish a threshold: the minimum percentage of pixels of a given color an object must have to receive that color label. We then compare these rules with their subsequent color attribution decisions. Our findings reveal that models systematically violate their own introspective rules. For example, GPT-5-mini violates its stated introspection rules in nearly 60\% of cases on objects with strong color priors. Human participants remain faithful to their stated rules, with any apparent violations being explained by a well-documented tendency to overestimate color coverage. In contrast, we find that VLMs are excellent estimators of color coverage, yet blatantly contradict their own reasoning in their final responses. Across all models and strategies for eliciting introspective rules, world-knowledge priors systematically degrade faithfulness in ways that do not mirror human cognition. Our findings challenge the view that VLM reasoning failures are difficulty-driven and suggest that VLM introspective self-knowledge is miscalibrated, with direct implications for high-stakes deployment.

This paper introduces Arabic-DeepSeek-R1, an application-driven open-source Arabic LLM that leverages a sparse MoE backbone to address the digital equity gap for under-represented languages, and establishes a new SOTA across the entire Open Arabic LLM Leaderboard (OALL). Our four-phase CoT distillation scheme integrates Arabic-specific linguistic verification and regional ethical norms into a 372M-token, contamination-controlled 80/20 Arabic-English training mixture. Arabic-DeepSeek-R1 achieves the highest average score across the seven-benchmark OALL suite while establishing SOTA or near-SOTA, including dominant results on grammar-focused MadinahQA (surpassing both GPT-5.1 and the OALL leader by substantial margins), safety-oriented AraTrust, multi-ability AlGhafa, and retrieval-augmented ALRAGE. Our results indicate that the combination of sparse MoE architecture, culturally-informed CoT distillation with explicit Arabic linguistic checks, and strategic bilingual data curation enables an open-source adapted model to systematically outperform the proprietary frontier system GPT-5.1 on the majority of benchmarks evaluating comprehensive language-specific tasks: the first such demonstration for Arabic LLMs. These findings indicate that much of Arabic's performance deficit in current LLM ecosystems stems from under-specialization rather than architectural limitations, and that parameter-efficient adaptation of open reasoning models can yield breakthrough SOTA performance without industrial-scale pretraining costs. Arabic-DeepSeek-R1 establishes a validated and replicable framework for sovereign and domain-specific language technologies, demonstrating that strategic, culturally-grounded adaptation of sparse MoE backbones offers a viable and cost-effective pathway to achieving record-breaking performance across standardized benchmarks for low-resource languages.

In recommender systems, collecting, storing, and processing large-scale interaction data is increasingly costly in terms of time, energy, and computation, yet it remains unclear when additional data stops providing meaningful gains. This paper investigates how offline recommendation performance evolves as the size of the training dataset increases and whether a saturation point can be observed. We implemented a reproducible Python evaluation workflow with two established toolkits, LensKit and RecBole, included 11 large public datasets with at least 7 million interactions, and evaluated 10 tool-algorithm combinations. Using absolute stratified user sampling, we trained models on nine sample sizes from 100,000 to 100,000,000 interactions and measured NDCG@10. Overall, raw NDCG usually increased with sample size, with no observable saturation point. To make result groups comparable, we applied min-max normalization within each group, revealing a clear positive trend in which around 75% of the points at the largest completed sample size also achieved the group's best observed performance. A late-stage slope analysis over the final 10-30% of each group further supported this upward trend: the interquartile range remained entirely non-negative with a median near 1.0. In summary, for traditional recommender systems on typical user-item interaction data, incorporating more training data remains primarily beneficial, while weaker scaling behavior is concentrated in atypical dataset cases and in the algorithmic outlier RecBole BPR under our setup.

Although LLM context lengths have grown, there is evidence that their ability to integrate information across long-form texts has not kept pace. We evaluate one such understanding task: generating summaries of novels. When human authors of summaries compress a story, they reveal what they consider narratively important. Therefore, by comparing human and LLM-authored summaries, we can assess whether models mirror human patterns of conceptual engagement with texts. To measure conceptual engagement, we align sentences from 150 human-written novel summaries with the specific chapters they reference. We demonstrate the difficulty of this alignment task, which indicates the complexity of summarization as a task. We then generate and align additional summaries by nine state-of-the-art LLMs for each of the 150 reference texts. Comparing the human and model-authored summaries, we find both stylistic differences between the texts and differences in how humans and LLMs distribute their focus throughout a narrative, with models emphasizing the ends of texts. Comparing human narrative engagement with model attention mechanisms suggests explanations for degraded narrative comprehension and targets for future development. We release our dataset to support future research.

Diffusion and flow matching models generate samples by learning time-dependent vector fields whose integration transports noise to data, requiring tens to hundreds of network evaluations at inference. We instead learn the transport map directly. We propose Optimal Transport Neural Flow Matching (OT-NFM), an ODE-free generative framework that parameterizes the flow map with neural flows, enabling true one-step generation with a single forward pass. We show that naive flow-map training suffers from mean collapse, where inconsistent noise-data pairings drive all outputs toward the data mean. We prove that consistent coupling is necessary for non-degenerate learning and address this using optimal transport pairings with scalable minibatch and online coupling strategies. Experiments on synthetic benchmarks and image generation tasks (MNIST and CIFAR-10) demonstrate competitive sample quality while reducing inference to a single network evaluation.

CubeSats have revolutionized access to space by providing affordable and accessible platforms for research and education. However, their reliance on Commercial Off-The-Shelf (COTS) components and open-source software has introduced significant cybersecurity vulnerabilities. Ensuring the cybersecurity of CubeSats is vital as they play increasingly important roles in space missions. Traditional security measures, such as intrusion detection systems (IDS), are impractical for CubeSats due to resource constraints and unique operational environments. This paper provides an in-depth review of current cybersecurity practices for CubeSats, highlighting limitations and identifying gaps in existing methods. Additionally, it explores non-cyber anomaly detection techniques that offer insights into adaptable algorithms and deployment strategies suitable for CubeSat constraints. Open research problems are identified, including the need for resource-efficient intrusion detection mechanisms, evaluation of IDS solutions under realistic mission scenarios, development of autonomous response systems, and creation of cybersecurity frameworks. The addition of TinyML into CubeSat systems is explored as a promising solution to address these challenges, offering resource-efficient, real-time intrusion detection capabilities. Future research directions are proposed, such as integrating cybersecurity with health monitoring systems, and fostering collaboration between cybersecurity researchers and space domain experts.

LLM agents increasingly draft messages on behalf of users, yet users routinely overshare sensitive information and disagree on what counts as private. Existing systems support only suppression (omitting sensitive information) and generalization (replacing information with an abstraction), and are typically evaluated on single isolated messages, leaving both the strategy space and evaluation setting incomplete. We formalize privacy-preserving LLM communication as an \textbf{Information Sufficiency (IS)} task, introduce \textbf{free-text pseudonymization} as a third strategy that replaces sensitive attributes with functionally equivalent alternatives, and propose a \textbf{conversational evaluation protocol} that assesses strategies under realistic multi-turn follow-up pressure. Across 792 scenarios spanning three power-relation types (institutional, peer, intimate) and three sensitivity categories (discrimination risk, social cost, boundary), we evaluate seven frontier LLMs on privacy at two granularities, covertness, and utility. Pseudonymization yields the strongest privacy\textendash utility tradeoff overall, and single-message evaluation systematically underestimates leakage, with generalization losing up to 16.3 percentage points of privacy under follow-up.

Data harmonization remains a major bottleneck for integrative analysis due to heterogeneity in schemas, value representations, and domain-specific conventions. BDI-Kit provides an extensible toolkit for schema and value matching. It exposes two complementary interfaces tailored to different user needs: a Python API enabling developers to construct harmonization pipelines programmatically, and an AI-assisted chat interface allowing domain experts to harmonize data through natural language dialogue. This demonstration showcases how users interact with BDI-Kit to iteratively explore, validate, and refine schema and value matches through a combination of automated matching, AI-assisted reasoning, and user-driven refinement. We present two scenarios: (i) using the Python API to programmatically compose primitives, examine intermediate outputs, and reuse transformations; and (ii) conversing with the AI assistant in natural language to access BDI-Kit's capabilities and iteratively refine outputs based on the assistant's suggestions.

The paper presents an approach for the recognition of toxic habits named entities in Spanish clinical texts. The approach was developed for the ToxHabits Shared Task. Our team participated in subtask 1, which aims to detect substance use and abuse mentions in clinical case reports and classify them in four categories (Tobacco, Alcohol, Cannabis, and Drug). We explored various methods of utilizing LLMs for the task, including zero-shot, few-shot, and prompt optimization, and found that GPT-4.1's few-shot prompting performed the best in our experiments. Our method achieved an F1 score of 0.65 on the test set, demonstrating a promising result for recognizing named entities in languages other than English.

The large language models (LLMs) might produce a persuasive argument within mathematical and logical fields, although such argument often includes some minor missteps, including the entire omission of side conditions, invalid inference patterns, or appeals to a lemma that cannot be derived logically out of the context being discussed. These omissions are infamously hard to notice solely out of the text, as even the misconstrued construction still may seem mostly accurate. Conversely, interactive theorem provers like Lean and Coq have rigorous reliability by ensuring that syntactic and semantic statements only accept statements that can pass all the syntactic and semantic steps in the program which is a small trusted kernel of the language type-checks with. Despite the fact that this technique provides strong guarantees, it comes at quite a heavy price: the evidence must be completely formalized, and the evidence user or a auxiliary search program must provide an avalanche of low-level information. This paper presents a hybrid pipeline where an LLM generates a typed proof sketch in a compact DSL and a lightweight trusted kernel expands the sketch into explicit proof obligations.

Global ocean models exhibit biases in the mean state and variability, particularly at coarse resolution, where mesoscale eddies are unresolved. To address these biases, parameterization coefficients are typically tuned ad hoc. Here, we formulate parameter tuning as a calibration problem using Ensemble Kalman Inversion (EKI). We optimize parameters of a neural network parameterization of mesoscale eddies in two idealized ocean models at coarse resolution. The calibrated parameterization reduces errors in the time-averaged fluid interfaces and their variability by approximately a factor of two compared to the unparameterized model or the offline-trained parameterization. The EKI method is robust to noise in time-averaged statistics arising from chaotic ocean dynamics. Furthermore, we propose an efficient calibration protocol that bypasses integration to statistical equilibrium by carefully choosing an initial condition. These results demonstrate that systematic calibration can substantially improve coarse-resolution ocean simulations and provide a practical pathway for reducing biases in global ocean models.

Spiking reservoir computing provides an energy-efficient approach to temporal processing, but reliably tuning reservoirs to operate at the edge-of-chaos is challenging due to experimental uncertainty. This work bridges abstract notions of criticality and practical stability by introducing and exploiting the robustness interval, an operational measure of the hyperparameter range over which a reservoir maintains performance above task-dependent thresholds. Through systematic evaluations of Leaky Integrate-and-Fire (LIF) architectures on both static (MNIST) and temporal (synthetic Ball Trajectories) tasks, we identify consistent monotonic trends in the robustness interval across a broad spectrum of network configurations: the robustness-interval width decreases with presynaptic connection density $β$ (i.e., directly with sparsity) and directly with the firing threshold $θ$. We further identify specific $(β, θ)$ pairs that preserve the analytical mean-field critical point $w_{\text{crit}}$, revealing iso-performance manifolds in the hyperparameter space. Control experiments on Erdős-Rényi graphs show the phenomena persist beyond small-world topologies. Finally, our results show that $w_{\text{crit}}$ consistently falls within empirical high-performance regions, validating $w_{\text{crit}}$ as a robust starting coordinate for parameter search and fine-tuning. To ensure reproducibility, the full Python code is publicly available.

Hallucination in large language models (LLMs) continues to be a significant issue, particularly in tasks like question answering, where models often generate plausible yet incorrect or irrelevant information. Although various methods have been proposed to mitigate hallucinations, the relationship between attention patterns and hallucinations has not been fully explored. In this paper, we analyze the distribution of attention scores across each layer and attention head of LLMs, revealing a common and intriguing phenomenon: shallow layers of LLMs primarily rely on uniform attention patterns, where the model distributes its attention evenly across the entire sequence. This uniform attention pattern can lead to hallucinations, as the model fails to focus on the most relevant information. To mitigate this issue, we propose a training-free method called Attention Replacement Technique (ART), which replaces these uniform attention patterns in the shallow layers with local attention patterns. This change directs the model to focus more on the relevant contexts, thus reducing hallucinations. Through extensive experiments, ART demonstrates significant reductions in hallucinations across multiple LLM architectures, proving its effectiveness and generalizability without requiring fine-tuning or additional training data.

We present Qualixar OS, the first application-layer operating system for universal AI agent orchestration. Unlike kernel-level approaches (AIOS) or single-framework tools (AutoGen, CrewAI), Qualixar OS provides a complete runtime for heterogeneous multi-agent systems spanning 10 LLM providers, 8+ agent frameworks, and 7 transports. We contribute: (1) execution semantics for 12 multi-agent topologies including grid, forest, mesh, and maker patterns; (2) Forge, an LLM-driven team design engine with historical strategy memory; (3) three-layer model routing combining Q-learning, five strategies, and Bayesian POMDP with dynamic multi-provider discovery; (4) a consensus-based judge pipeline with Goodhart detection, JSD drift monitoring, and alignment trilemma navigation; (5) four-layer content attribution with HMAC signing and steganographic watermarks; (6) universal compatibility via the Claw Bridge supporting MCP and A2A protocols with a 25-command Universal Command Protocol; (7) a 24-tab production dashboard with visual workflow builder and skill marketplace. Qualixar OS is validated by 2,821 test cases across 217 event types and 8 quality modules. On a custom 20-task evaluation suite, the system achieves 100% accuracy at a mean cost of $0.000039 per task. Source-available under the Elastic License 2.0.

Graphs are a central representation in biomedical research, capturing molecular interaction networks, gene regulatory circuits, cell--cell communication maps, and knowledge graphs. Despite their importance, currently there is not a broadly reusable foundation model available for graph analysis comparable to those that have transformed language and vision. Existing graph neural networks are typically trained on a single dataset and learn representations specific only to that graph's node features, topology, and label space, limiting their ability to transfer across domains. This lack of generalization is particularly problematic in biology and medicine, where networks vary substantially across cohorts, assays, and institutions. Here we introduce a graph foundation model designed to learn transferable structural representations that are not specific to specific node identities or feature schemes. Our approach leverages feature-agnostic graph properties, including degree statistics, centrality measures, community structure indicators, and diffusion-based signatures, and encodes them as structural prompts. These prompts are integrated with a message-passing backbone to embed diverse graphs into a shared representation space. The model is pretrained once on heterogeneous graphs and subsequently reused on unseen datasets with minimal adaptation. Across multiple benchmarks, our pretrained model matches or exceeds strong supervised baselines while demonstrating superior zero-shot and few-shot generalization on held-out graphs. On the SagePPI benchmark, supervised fine-tuning of the pretrained backbone achieves a mean ROC-AUC of 95.5%, a gain of 21.8% over the best supervised message-passing baseline. The proposed technique thus provides a unique approach toward reusable, foundation-scale models for graph-structured data in biomedical and network science applications.

Background: Colorectal cancer (CRC) remains a leading cause of cancer-related mortality worldwide. Accurate survival prediction is essential for treatment stratification, yet existing pathology foundation models often overlook organ-specific features critical for CRC prognostication. Methods: We propose MorphDistill, a two-stage framework that distills complementary knowledge from multiple pathology foundation models into a compact CRC-specific encoder. In Stage I, a student encoder is trained using dimension-agnostic multi-teacher relational distillation with supervised contrastive regularization on large-scale colorectal datasets. This preserves inter-sample relationships from ten foundation models without explicit feature alignment. In Stage II, the encoder extracts patch-level features from whole-slide images, which are aggregated via attention-based multiple instance learning to predict five-year survival. Results: On the Alliance/CALGB 89803 cohort (n=424, stage III CRC), MorphDistill achieves an AUC of 0.68 (SD 0.08), an approximately 8% relative improvement over the strongest baseline (AUC 0.63). It also attains a C-index of 0.661 and a hazard ratio of 2.52 (95% CI: 1.73-3.65), outperforming all baselines. On an external TCGA cohort (n=562), it achieves a C-index of 0.628, demonstrating strong generalization across datasets and robustness across clinical subgroups. Conclusion: MorphDistill enables task-specific representation learning by integrating knowledge from multiple foundation models into a unified encoder. This approach provides an efficient strategy for prognostic modeling in computational pathology, with potential for broader oncology applications. Further validation across additional cohorts and disease stages is warranted.

Uncertainty estimation for reasoning language models remains difficult to deploy in practice: sampling-based methods are computationally expensive, while common single-pass proxies such as verbalized confidence or trace length are often inconsistent across models. This problem is compounded for proprietary reasoning APIs that expose neither logits nor intermediate token probabilities, leaving practitioners with no reliable uncertainty signal at inference time. We propose SELFDOUBT, a single-pass uncertainty framework that resolves this impasse by extracting behavioral signals directly from the reasoning trace itself. Our key signal, the Hedge-to-Verify Ratio (HVR), detects whether a reasoning trace contains uncertainty markers and, if so, whether they are offset by explicit selfchecking behavior. Unlike methods that require multiple sampled traces or model internals, SELFDOUBT operates on a single observed reasoning trajectory, making it suitable for latency- and cost-constrained deployment over any proprietary API. We evaluate SELFDOUBT across seven models and three multi-step reasoning benchmarks (BBH, GPQA-Diamond, and MMLU-Pro). Most notably, traces containing no hedging markers are correct 96% of the time, revealing an emergent high-precision confidence gate at zero additional cost. For the remaining cases, the full SELFDOUBT score significantly outperforms sampling-based semantic entropy at 10x lower inference cost. A deployment cascade combining both stages attains 90% accuracy at 71% coverage without any task-specific labels. These results establish SELFDOUBT as a scalable, production-ready foundation for uncertainty estimation over proprietary reasoning models.

Accurate reconstruction of environmental scalar fields from sparse onboard observations is essential for autonomous vehicles engaged in aquatic monitoring. Beyond point estimates, principled uncertainty quantification is critical for active sensing strategies such as Informative Path Planning, where epistemic uncertainty drives data collection decisions. This paper compares Gaussian Processes, Monte Carlo Dropout, Deep Ensembles, and Evidential Deep Learning for simultaneous scalar field reconstruction and uncertainty decomposition under three perceptual models representative of real sensor modalities. Results show that Evidential Deep Learning achieves the best reconstruction accuracy and uncertainty calibration across all sensor configurations at the lowest inference cost, while Gaussian Processes are fundamentally limited by their stationary kernel assumption and become intractable as observation density grows. These findings support Evidential Deep Learning as the preferred method for uncertainty-aware field reconstruction in real-time autonomous vehicle deployments.

We present an innovative multi-stage optimization strategy combining reinforcement learning (RL) and supervised fine-tuning (SFT) to enhance the pedagogical knowledge of large language models (LLMs), as illustrated by EduQwen 32B-RL1, EduQwen 32B-SFT, and an optional third-stage model EduQwen 32B-SFT-RL2: (1) RL optimization that implements progressive difficulty training, focuses on challenging examples, and employs extended reasoning rollouts; (2) a subsequent SFT phase that leverages the RL-trained model to synthesize high-quality training data with difficulty-weighted sampling; and (3) an optional second round of RL optimization. EduQwen 32B-RL1, EduQwen 32B-SFT, and EduQwen 32B-SFT-RL2 are an application-driven family of open-source pedagogical LLMs built on a dense Qwen3-32B backbone. These models remarkably achieve high enough accuracy on the Cross-Domain Pedagogical Knowledge (CDPK) Benchmark to establish new state-of-the-art (SOTA) results across the interactive Pedagogy Benchmark Leaderboard and surpass significantly larger proprietary systems such as the previous benchmark leader Gemini-3 Pro. These dense 32-billion-parameter models demonstrate that domain-specialized optimization can transform mid-sized open-source LLMs into true pedagogical domain experts that outperform much larger general-purpose systems, while preserving the transparency, customizability, and cost-efficiency required for responsible educational AI deployment.

In many real-world settings, institutions can and do adjust the consequences attached to algorithmic classification decisions, such as the size of fines, sentence lengths, or benefit levels. We refer to these consequences as the stakes associated with classification. These stakes can give rise to behavioral responses to classification, as people adjust their actions in anticipation of how they will be classified. Much of the algorithmic fairness literature evaluates classification outcomes while holding behavior fixed, treating behavioral differences across groups as exogenous features of the environment. Under this assumption, the stakes of classification play no role in shaping outcomes. We revisit classic impossibility results in algorithmic fairness in a setting where people respond strategically to classification. We show that, in this environment, the well-known incompatibility between error-rate balance and predictive parity disappears, but only by potentially introducing a qualitatively different form of unequal treatment. Concretely, we construct a two-stage design in which a classifier first standardizes its statistical performance across groups, and then adjusts stakes so as to induce comparable patterns of behavior. This requires treating groups differently in the consequences attached to identical classification decisions. Our results demonstrate that fairness in strategic settings cannot be assessed solely by how algorithms map data into decisions. Rather, our analysis treats the human consequences of classification as primary design variables, introduces normative criteria governing their use, and shows that their interaction with statistical fairness criteria generates qualitatively new tradeoffs. Our aim is to make these tradeoffs precise and explicit.

We investigate whether post-trained capabilities can be transferred across models without retraining, with a focus on transfer across different model scales. We propose the Master Key Hypothesis, which states that model capabilities correspond to directions in a low-dimensional latent subspace that induce specific behaviors and are transferable across models through linear alignment. Based on this hypothesis, we introduce UNLOCK, a training-free and label-free framework that extracts a capability direction by contrasting activations between capability-present and capability-absent Source variants, aligns it with a Target model through a low-rank linear transformation, and applies it at inference time to elicit the behavior. Experiments on reasoning behaviors, including Chain-of-Thought (CoT) and mathematical reasoning, demonstrate substantial improvements across model scales without training. For example, transferring CoT reasoning from Qwen1.5-14B to Qwen1.5-7B yields an accuracy gain of 12.1% on MATH, and transferring a mathematical reasoning direction from Qwen3-4B-Base to Qwen3-14B-Base improves AGIEval Math accuracy from 61.1% to 71.3%, surpassing the 67.8% achieved by the 14B post-trained model. Our analysis shows that the success of transfer depends on the capabilities learned during pre-training, and that our intervention amplifies latent capabilities by sharpening the output distribution toward successful reasoning trajectories.

AI-driven symptom analysis systems face persistent challenges in reliability, interpretability, and hallucination. End-to-end generative approaches often lack traceability and may produce unsupported or inconsistent diagnostic outputs in safety-critical settings. We present SymptomWise, a framework that separates language understanding from diagnostic reasoning. The system combines expert-curated medical knowledge, deterministic codex-driven inference, and constrained use of large language models. Free-text input is mapped to validated symptom representations, then evaluated by a deterministic reasoning module operating over a finite hypothesis space to produce a ranked differential diagnosis. Language models are used only for symptom extraction and optional explanation, not for diagnostic inference. This architecture improves traceability, reduces unsupported conclusions, and enables modular evaluation of system components. Preliminary evaluation on 42 expert-authored challenging pediatric neurology cases shows meaningful overlap with clinician consensus, with the correct diagnosis appearing in the top five differentials in 88% of cases. Beyond medicine, the framework generalizes to other abductive reasoning domains and may serve as a deterministic structuring and routing layer for foundation models, improving precision and potentially reducing unnecessary computational overhead in bounded tasks.

Latent reasoning via continuous chain-of-thoughts (Latent CoT) has emerged as a promising alternative to discrete CoT reasoning. Operating in continuous space increases expressivity and has been hypothesized to enable superposition: the ability to maintain multiple candidate solutions simultaneously within a single representation. Despite theoretical arguments, it remains unclear whether language models actually leverage superposition when reasoning using latent CoTs. We investigate this question across three regimes: a training-free regime that constructs latent thoughts as convex combinations of token embeddings, a fine-tuned regime where a base model is adapted to produce latent thoughts, and a from-scratch regime where a model is trained entirely with latent thoughts to solve a given task. Using Logit Lens and entity-level probing to analyze internal representations, we find that only models trained from scratch exhibit signs of using superposition. In the training-free and fine-tuned regimes, we find that the superposition either collapses or is not used at all, with models discovering shortcut solutions instead. We argue that this is due to two complementary phenomena: i) pretraining on natural language data biases models to commit to a token in the last layers ii) capacity has a huge effect on which solutions a model favors. Together, our results offer a unified explanation for when and why superposition arises in continuous chain-of-thought reasoning, and identify the conditions under which it collapses.

New generation of AI coding tools, including AI-powered IDEs equipped with agentic capabilities, can generate code within the context of the project. These AI IDEs are increasingly perceived as capable of producing project-level code at scale. However, there is limited empirical evidence on the extent to which they can generate large-scale software systems and what design issues such systems may exhibit. To address this gap, we conducted a study to explore the capability of Cursor in generating large-scale projects and to evaluate the design quality of projects generated by Cursor. First, we propose a Feature-Driven Human-In-The-Loop (FD-HITL) framework that systematically guides project generation from curated project descriptions. We generated 10 projects using Cursor with the FD-HITL framework across three application domains and multiple technologies. We assessed the functional correctness of these projects through manual evaluation, obtaining an average functional correctness score of 91%. Next, we analyzed the generated projects using two static analysis tools, CodeScene and SonarQube, to detect design issues. We identified 1,305 design issues categorized into 9 categories by CodeScene and 3,193 issues in 11 categories by SonarQube. Our findings show that (1) when used with the FD-HITL framework, Cursor can generate functional large-scale projects averaging 16,965 LoC and 114 files; (2) the generated projects nevertheless contain design issues that may pose long-term maintainability and evolvability risks, requiring careful review by experienced developers; (3) the most prevalent issues include Code Duplication, high Code Complexity, Large Methods, Framework Best-Practice Violations, Exception-Handling Issues and Accessibility Issues; (4) these design issues violate design principles such as SRP, SoC, and DRY. The replication package is at https://github.com/Kashifraz/DIinAGP

The serving paradigm of large language models (LLMs) is rapidly shifting towards complex multi-agent workflows where specialized agents collaborate over massive shared contexts. While Low-Rank Adaptation (LoRA) enables the efficient co-hosting of these specialized agents on a single base model, it introduces a critical memory footprint bottleneck during serving. Specifically, unique LoRA activations cause Key-Value (KV) cache divergence across agents, rendering traditional prefix caching ineffective for shared contexts. This forces redundant KV cache maintenance, rapidly saturating GPU capacity and degrading throughput. To address this challenge, we introduce ForkKV, a serving system for multi-LoRA agent workflows centered around a novel memory management paradigm in OS: fork with copy-on-write (CoW). By exploiting the structural properties of LoRA, ForkKV physically decouples the KV cache into a massive shared component (analogous to the parent process's memory pages) and lightweight agent-specific components (the child process's pages). To support this mechanism, we propose a DualRadixTree architecture that allows newly forked agents to inherit the massive shared cache and apply CoW semantics for their lightweight unique cache. Furthermore, to guarantee efficient execution, we design ResidualAttention, a specialized kernel that reconstructs the disaggregated KV cache directly within on-chip SRAM. Comprehensive evaluations across diverse language models and practical datasets of different tasks demonstrate that ForkKV achieves up to 3.0x the throughput of state-of-the-art multi-LoRA serving systems with a negligible impact on generation quality.

Web agents automate browser tasks, ranging from simple form completion to complex workflows like ordering groceries. While current benchmarks evaluate general-purpose performance~(e.g., WebArena) or safety against malicious actions~(e.g., SafeArena), no existing framework assesses an agent's ability to successfully execute user-facing website security and privacy tasks, such as managing cookie preferences, configuring privacy-sensitive account settings, or revoking inactive sessions. To address this gap, we introduce WebSP-Eval, an evaluation framework for measuring web agent performance on website security and privacy tasks. WebSP-Eval comprises 1) a manually crafted task dataset of 200 task instances across 28 websites; 2) a robust agentic system supporting account and initial state management across runs using a custom Google Chrome extension; and 3) an automated evaluator. We evaluate a total of 8 web agent instantiations using state-of-the-art multimodal large language models, conducting a fine-grained analysis across websites, task categories, and UI elements. Our evaluation reveals that current models suffer from limited autonomous exploration capabilities to reliably solve website security and privacy tasks, and struggle with specific task categories and websites. Crucially, we identify stateful UI elements such as toggles and checkboxes are a primary reason for agent failure, failing at a rate of more than 45\% in tasks containing these elements across many models.

Deep linear networks (DLNs) are used as an analytically tractable model of the training dynamics of deep neural networks. While gradient descent in DLNs is known to exhibit saddle-to-saddle dynamics, the impact of stochastic gradient descent (SGD) noise on this regime remains poorly understood. We investigate the dynamics of SGD during training of DLNs in the saddle-to-saddle regime. We model the training dynamics as stochastic Langevin dynamics with anisotropic, state-dependent noise. Under the assumption of aligned and balanced weights, we derive an exact decomposition of the dynamics into a system of one-dimensional per-mode stochastic differential equations. This establishes that the maximal diffusion along a mode precedes the corresponding feature being completely learned. We also derive the stationary distribution of SGD for each mode: in the absence of label noise, its marginal distribution along specific features coincides with the stationary distribution of gradient flow, while in the presence of label noise it approximates a Boltzmann distribution. Finally, we confirm experimentally that the theoretical results hold qualitatively even without aligned or balanced weights. These results establish that SGD noise encodes information about the progression of feature learning but does not fundamentally alter the saddle-to-saddle dynamics.

Arabic medical text generation is increasingly needed to help users interpret symptoms and access general health guidance in their native language. Nevertheless, many existing methods assume uniform importance across training samples, overlooking differences in clinical severity. This simplification can hinder the model's ability to properly capture complex or high-risk cases. To overcome this issue, this work introduces a Severity-based Curriculum Learning Strategy for Arabic Medical Text Generation, where the training process is structured to move gradually from less severe to more critical medical conditions. The approach divides the dataset into ordered stages based on severity and incrementally exposes the model to more challenging cases during fine-tuning, allowing it to first learn basic medical patterns before addressing more complex scenarios. The proposed method is evaluated on a subset of the Medical Arabic Question Answering (MAQA) dataset, which includes Arabic medical questions describing symptoms alongside corresponding responses. In addition, the dataset is annotated with three severity levels (Mild, Moderate, and Critical) using a rule-based method developed in this study. The results demonstrate that incorporating severity-aware curriculum learning leads to consistent performance improvements across all tested models, with gains of around +4% to +7% over baseline models and +3% to +6% compared with conventional fine-tuning approaches.

The adoption of Machine and Deep Learning (ML/DL) technologies introduces maintenance challenges, leading to Technical Debt (TD). Algorithm Debt (AD) is a TD type that impacts the performance and scalability of ML/DL systems. A review of 42 primary studies expanded AD's definition, uncovered its implicit presence, identified its smells, and highlighted future directions. These findings will guide an AD-focused study, enhancing the reliability of ML/DL systems.

Exploring ensemble simulations is increasingly important across many scientific domains. However, supporting flexible post-hoc exploration remains challenging due to the trade-off between storing the expensive raw data and flexibly adjusting visualization settings. Existing visualization surrogate models have improved this workflow, but they either operate in image space without an explicit 3D representation or rely on neural radiance fields that are computationally expensive for interactive exploration and encode all parameter-driven variations within a single implicit field. In this work, we introduce GS-Surrogate, a deformable Gaussian Splatting-based visualization surrogate for parameter-space exploration. Our method first constructs a canonical Gaussian field as a base 3D representation and adapts it through sequential parameter-conditioned deformations. By separating simulation-related variations from visualization-specific changes, this explicit formulation enables efficient and controllable adaptation to different visualization tasks, such as isosurface extraction and transfer function editing. We evaluate our framework on a range of simulation datasets, demonstrating that GS-Surrogate enables real-time and flexible exploration across both simulation and visualization parameter spaces.

In-Context Learning (ICL) has been extensively studied in text-only Language Models, but remains largely unexplored in the speech domain. Here, we investigate how linguistic and acoustic features affect ICL in Speech Language Models. We focus on the Text-to-Speech (TTS) task, which allows us to analyze ICL from two angles: (1) how accurately the model infers the task from the demonstrations (i.e., generating the correct spoken content), and (2) to what extent the model mimics the acoustic characteristics of the demonstration speech in its output. We find that speaking rate strongly affects ICL performance and is also mimicked in the output, whereas pitch range and intensity have little impact on performance and are not consistently reproduced. Finally, we investigate the role of induction heads in speech-based ICL and show that these heads play a causal role: ablating the top-k induction heads completely removes the model's ICL ability, mirroring findings from text-based ICL.

Accurate dietary assessment is critical for precision nutrition, yet most image-based methods rely on a single pre-consumption image and provide only coarse, meal-level estimates. These approaches cannot determine what was actually consumed and often require restrictive inputs such as depth sensing, multi-view imagery, or explicit segmentation. In this paper, we propose a simple vision-language framework for food-item-level nutritional analysis using paired before-and-after eating images. Instead of relying on rigid segmentation masks, our method leverages natural language prompts to localize specific food items and estimate their weight directly from a single RGB image. We further estimate food consumption by predicting weight differences between paired images using a two-stage training strategy. We evaluate our method on three publicly available datasets and demonstrate consistent improvements over existing approaches, establishing a strong baseline for before-and-after dietary image analysis.

Generalizing from a single labeled source domain to unseen target domains, without access to any target data during training, remains a fundamental challenge in robust machine learning. We address this underexplored setting, known as Single Domain Generalization (SDG), by proposing BiSDG, a bi-level optimization framework that explicitly decouples task learning from domain modeling. BiSDG simulates distribution shifts through surrogate domains constructed via label-preserving transformations of the source data. To capture domain-specific context, we propose a domain prompt encoder that generates lightweight modulation signals to produce augmenting features via feature-wise linear modulation. The learning process is formulated as a bi-level optimization problem: the inner objective optimizes task performance under fixed prompts, while the outer objective maximizes generalization across the surrogate domains by updating the domain prompt encoder. We further develop a practical gradient approximation scheme that enables efficient bi-level training without second-order derivatives. Extensive experiments on various SGD benchmarks demonstrate that BiSDG consistently outperforms prior methods, setting new state-of-the-art performance in the SDG setting.

Large language models have shown strong potential for Arabic medical text generation; however, traditional fine-tuning objectives treat all medical cases uniformly, ignoring differences in clinical severity. This limitation is particularly critical in healthcare settings, where errors in severe cases contain higher clinical risk. In this work, we propose a severity-aware weighted loss for fine-tuning Arabic language models on medical complaint-response data. The method depends on soft severity probabilities to dynamically scale token-level loss contributions during optimization, thereby prioritizing clinically critical interactions without modifying model architectures. Experiments are conducted using the MAQA dataset, which provides Arabic medical complaints and trusted human responses. Severity labels and probabilistic scores are automatically derived using a fine-tuned AraBERT-based classifier and incorporated exclusively at the loss level. The proposed approach is evaluated across ten Arabic large language models of varying architectures and parameter scales. While standard cross-entropy fine-tuning yields only modest improvements, severity-aware optimization consistently achieves larger gains. Using a balanced weighting configuration, performance improves from 54.04% to 66.14% for AraGPT2-Base, from 59.16% to 67.18% for AraGPT2-Medium, and from 57.83% to 66.86% for Qwen2.5-0.5B, with peak performance reaching 67.18%. Overall, severity-aware fine-tuning delivers improvements of up to 12.10% over non-fine-tuned baselines, demonstrating robust and architecture-consistent gains.

Although tension between university curricula and industry expectations has existed in some form for decades, the rapid integration of generative AI (GenAI) tools into software development has recently widened the gap between the two domains. To better understand this disconnect, we surveyed 51 industry practitioners (software developers, technical leads, upper management, \etc) and conducted 11 follow-up interviews focused on hiring practices, required job skills, perceived shortcomings in university curricula, and views on how university learning outcomes can be improved. Our results suggest that GenAI creates demand for new skills (\eg prompting and output evaluation), while strengthening the importance of soft-skills (\eg problem solving and critical thinking) and traditional competencies (\eg architecture design and debugging). We synthesize these findings into actionable recommendations for academia (\eg how to incorporate GenAI into curricula and evaluation redesign). Our work offers empirical guidance to help educators prepare students for modern software engineering environments.

Graph Transformers have recently attracted attention for molecular property prediction by combining the inductive biases of graph neural networks (GNNs) with the global receptive field of Transformers. However, many existing hybrid architectures remain GNN-dominated, causing the resulting representations to remain heavily shaped by local message passing. Moreover, most existing methods operate at only a single structural granularity, limiting their ability to capture molecular patterns that span multiple molecular scales. We introduce BiScale-GTR, a unified framework for self-supervised molecular representation learning that combines chemically grounded fragment tokenization with adaptive multi-scale reasoning. Our method improves graph Byte Pair Encoding (BPE) tokenization to produce consistent, chemically valid, and high-coverage fragment tokens, which are used as fragment-level inputs to a parallel GNN-Transformer architecture. Architecturally, atom-level representations learned by a GNN are pooled into fragment-level embeddings and fused with fragment token embeddings before Transformer reasoning, enabling the model to jointly capture local chemical environments, substructure-level motifs, and long-range molecular dependencies. Experiments on MoleculeNet, PharmaBench, and the Long Range Graph Benchmark (LRGB) demonstrate state-of-the-art performance across both classification and regression tasks. Attribution analysis further shows that BiScale-GTR highlights chemically meaningful functional motifs, providing interpretable links between molecular structure and predicted properties. Code will be released upon acceptance.

Drifting models generate high-quality samples in a single forward pass by transporting generated samples toward the data distribution using a vector valued drift field. We investigate whether this procedure is equivalent to optimizing a scalar loss and find that, in general, it is not: drift fields are not conservative - they cannot be written as the gradient of any scalar potential. We identify the position-dependent normalization as the source of non-conservatism. The Gaussian kernel is the unique exception where the normalization is harmless and the drift field is exactly the gradient of a scalar function. Generalizing this, we propose an alternative normalization via a related kernel (the sharp kernel) which restores conservatism for any radial kernel, yielding well-defined loss functions for training drifting models. While we identify that the drifting field matching objective is strictly more general than loss minimization, as it can implement non-conservative transport fields that no scalar loss can reproduce, we observe that practical gains obtained utilizing this flexibility are minimal. We thus propose to train drifting models with the conceptually simpler formulations utilizing loss functions.

Autonomous highway driving, especially for long-haul heavy trucks, requires detecting objects at long ranges beyond 500 meters to satisfy braking distance requirements at high speeds. At long distances, vehicles and other critical objects occupy only a few pixels in high-resolution images, causing state-of-the-art object detectors to fail. This challenge is compounded by the limited effective range of commercially available LiDAR sensors, which fall short of ultra-long range thresholds because of quadratic loss of resolution with distance, making image-based detection the most practically scalable solution given commercially available sensor constraints. We introduce Telescope, a two-stage detection model designed for ultra-long range autonomous driving. Alongside a powerful detection backbone, this model contains a novel re-sampling layer and image transformation to address the fundamental challenges of detecting small, distant objects. Telescope achieves $76\%$ relative improvement in mAP in ultra-long range detection compared to state-of-the-art methods (improving from an absolute mAP of 0.185 to 0.326 at distances beyond 250 meters), requires minimal computational overhead, and maintains strong performance across all detection ranges.

Diffusion Large Language Models (dLLMs) have achieved rapid progress, viewed as a promising alternative to the autoregressive paradigm. However, most dLLM decoders still adopt a global confidence threshold, and do not explicitly model local context from neighboring decoded states or temporal consistency of predicted token IDs across steps. To address this issue, we propose a simple spatio-temporal stability guided decoding approach, named STDec. We observe strong spatio-temporal stability in dLLM decoding: newly decoded tokens tend to lie near decoded neighbors, and their predicted IDs often remain consistent across several denoising steps. Inspired by this stability, our STDec includes spatial-aware decoding and temporal-aware decoding. The spatial-aware decoding dynamically generates the token-adaptive threshold by aggregating the decoded states of nearby tokens. The temporal-aware decoding relaxes the decoding thresholds for tokens whose predicted token IDs remain consistent over denoising steps. Our STDec is training-free and remains compatible with cache-based acceleration methods. Across textual reasoning and multimodal understanding benchmarks, STDec substantially improves throughput while maintaining comparable task performance score. Notably, on MBPP with LLaDA, STDec achieves up to 14.17x speedup with a comparable score. Homepage: https://yzchen02.github.io/STDec.

Recent diffusion-based text-to-speech (TTS) models achieve high naturalness and expressiveness, yet often suffer from speaker drift, a subtle, gradual shift in perceived speaker identity within a single utterance. This underexplored phenomenon undermines the coherence of synthetic speech, especially in long-form or interactive settings. We introduce the first automatic framework for detecting speaker drift by formulating it as a binary classification task over utterance-level speaker consistency. Our method computes cosine similarity across overlapping segments of synthesized speech and prompts large language models (LLMs) with structured representations to assess drift. We provide theoretical guarantees for cosine-based drift detection and demonstrate that speaker embeddings exhibit meaningful geometric clustering on the unit sphere. To support evaluation, we construct a high-quality synthetic benchmark with human-validated speaker drift annotations. Experiments with multiple state-of-the-art LLMs confirm the viability of this embedding-to-reasoning pipeline. Our work establishes speaker drift as a standalone research problem and bridges geometric signal analysis with LLM-based perceptual reasoning in modern TTS.

The rapid evolution of intelligent networks under the Internet of Everything (IoE) paradigm is transforming connectivity by integrating people, processes, data, and things. This ecosystem includes domains such as the Internet of Things (IoT), Internet of Healthcare (IoH), Internet of Vehicles (IoV), and cyber-physical and human-machine systems. While enabling efficiency and automation, this interconnectivity also exposes critical infrastructures to increasingly sophisticated cyber threats, creating an urgent need for advanced security solutions. This chapter examines the integration of Blockchain and Artificial Intelligence (AI) as complementary approaches for securing intelligent networks. Blockchain provides decentralized, immutable, and transparent mechanisms that strengthen data integrity, trust, and accountability. In parallel, AI offers predictive analytics, anomaly detection, and adaptive defense capabilities to enable proactive threat identification and mitigation. The chapter discusses how Blockchain supports security in cyber-physical systems, how AI enables proactive security operations, and how their combination creates robust, adaptive, and trustworthy security frameworks. The chapter also explores the emerging role of large language models in threat intelligence and analyzes how controlled agentic AI can support bounded security workflows such as alert triage, evidence collection, and policy-aware response planning. Representative case studies illustrate the potential of these technologies to enhance cyber resilience. Finally, challenges related to scalability, energy efficiency, and ethical considerations are addressed, along with reported mitigation strategies and future research directions. Overall, this chapter provides researchers, practitioners, and policymakers with insights to design secure, resilient, and adaptable intelligent networks.

Recent alignment work on Large Language Models (LLMs) suggests preference optimization can improve reasoning by shifting probability mass toward better solutions. We test this claim in a resource-constrained setting by applying GRPO with LoRA to SLMs (up to 3B) for math reasoning on GSM8K and MATH datasets with difficulty-stratified analyses. As problem difficulty increases, accuracy plateaus, revealing a capacity boundary: GRPO primarily reshapes output preferences without reliably improving hardest-tier solving. Consistent with this, training GRPO only on lower-difficulty problems matches full-dataset accuracy across difficulty tiers while using only ~45% training steps, indicating diminishing returns from harder samples in this regime. We also find a cross-dataset generalization effect: GSM8K-trained GRPO achieves higher accuracy on the numeric subset of MATH than MATH-trained GRPO, exceeding it by ~5% at 1.5B and by ~3% at 3B. We show that the best achievable gains depend strongly on the base model's prior reasoning competence and the dataset's difficulty profile.

Given the growing reliance on private data in training Large Language Models (LLMs), Federated Learning (FL) combined with Parameter-Efficient Fine-Tuning (PEFT) has garnered significant attention for enhancing privacy and efficiency. Despite FL's privacy benefits, prior studies have shown that private data can still be extracted from shared gradients. However, these studies, mainly on full-parameter model training, are limited to reconstructing small batches, short input sequences, and specific model architectures, such as encoder-based or decoder-based models. The reconstruction quality becomes even worse when dealing with gradients from PEFT methods. To fully understand the practical attack surface of federated LLMs, this paper proposes FedSpy-LLM, a scalable and generalizable data reconstruction attack designed to reconstruct training data with larger batch sizes and longer sequences while generalizing across diverse model architectures, even when PEFT methods are deployed for training. At the core of FedSpy-LLM is a novel gradient decomposition strategy that exploits the rank deficiency and subspace structure of gradients, enabling efficient token extraction while preserving key signal components at scale. This approach further mitigates the reconstruction challenges introduced by PEFT's substantial null space, ensuring robustness across encoder-based, decoder-based, and encoder-decoder model architectures. Additionally, by iteratively aligning each token's partial-sequence gradient with the full-sequence gradient, FedSpy-LLM ensures accurate token ordering in reconstructed sequences.

AI agents are increasingly deployed in real-world applications, including systems such as Manus, OpenClaw, and coding agents. Existing research has primarily focused on \emph{server-side} efficiency, proposing methods such as caching, speculative execution, traffic scheduling, and load balancing to reduce the cost of serving agentic workloads. However, as users increasingly construct agents by composing local tools, remote APIs, and diverse models, an equally important optimization problem arises on the client side. Client-side optimization asks how developers should allocate the resources available to them, including model choice, local tools, and API budget across pipeline stages, subject to application-specific quality, cost, and latency constraints. Because these objectives depend on the task and deployment setting, they cannot be determined by server-side systems alone. We introduce AgentOpt, the first framework-agnostic Python package for client-side agent optimization. We first study model selection, a high-impact optimization lever in multi-step agent pipelines. Given a pipeline and a small evaluation set, the goal is to find the most cost-effective assignment of models to pipeline roles. This problem is consequential in practice: at matched accuracy, the cost gap between the best and worst model combinations can reach 13--32$\times$ in our experiments. To efficiently explore the exponentially growing combination space, AgentOpt implements eight search algorithms, including Arm Elimination, Epsilon-LUCB, Threshold Successive Elimination, and Bayesian Optimization. Across four benchmarks, Arm Elimination recovers near-optimal accuracy while reducing evaluation budget by 24--67\% relative to brute-force search on three of four tasks. Code and benchmark results available at https://agentoptimizer.github.io/agentopt/.

Legal information retrieval in Portuguese remains difficult to evaluate systematically because available datasets differ widely in document type, query style, and relevance definition. We present JUÁ, a public benchmark for Brazilian legal retrieval designed to support more reproducible and comparable evaluation across heterogeneous legal collections. More broadly, JUÁ is intended not only as a benchmark, but as a continuous evaluation infrastructure for Brazilian legal IR, combining shared protocols, common ranking metrics, fixed splits when applicable, and a public leaderboard. The benchmark covers jurisprudence retrieval as well as broader legislative, regulatory, and question-driven legal search. We evaluate lexical, dense, and BM25-based reranking pipelines, including a domain-adapted Qwen embedding model fine-tuned on JUÁ-aligned supervision. Results show that the benchmark is sufficiently heterogeneous to distinguish retrieval paradigms and reveal substantial cross-dataset trade-offs. Domain adaptation yields its clearest gains on the supervision-aligned JUÁ-Juris subset, while BM25 remains highly competitive on other collections, especially in settings with strong lexical and institutional phrasing cues. Overall, JUÁ provides a practical evaluation framework for studying legal retrieval across multiple Brazilian legal domains under a common benchmark design.

Video streaming analytics is a crucial workload for vision-language model serving, but the high cost of multimodal inference limits scalability. Prior systems reduce inference cost by exploiting temporal and spatial redundancy in video streams, but they target either the vision transformer (ViT) or the LLM with a limited view, leaving end-to-end opportunities untapped. Moreover, existing methods incur significant overhead to identify redundancy, either through offline profiling and training or costly online computation, making them ill-suited for dynamic real-time streams. We present CodecFlow, a codec-guided streaming video analytics system built on a key observation that video codecs already extract the temporal and spatial structure of each stream as a byproduct of compression. CodecFlow treats this codec metadata as a low-cost runtime signal to unify optimization across video decoding, visual processing, and LLM prefilling, with transmission reduction as an inherent benefit of operating directly on compressed bitstreams. This drives codec-guided patch pruning before ViT encoding and selective key-value cache refresh during LLM prefilling, both of which are fully online and do not require offline training. Experiments show that CodecFlow achieves up to 3x throughput improvement and up to 87% GPU compute reduction over state-of-the-art baselines, while maintaining competitive accuracy with only 0-8% F1 drop.

This study examines the perception of legal professionals on the governance of AI in developing countries, using Nigeria as a case study. The study focused on ethical risks, regulatory gaps, and institutional readiness. The study adopted a qualitative case study design. Data were collected through 27 semi-structured interviews with legal practitioners in Nigeria. A focus group discussion was also held with seven additional legal practitioners across sectors such as finance, insurance, and corporate law. Thematic analysis was employed to identify key patterns in participant responses. Findings showed that there were concerns about data privacy risks and the lack of enforceable legal frameworks. Participants expressed limited confidence in institutional capacity and emphasized the need for locally adapted governance models rather than direct adoption of foreign frameworks. While some expressed optimism about AI's potential, this was conditional on the presence of strong legal oversight and public accountability. The study contributes to the growing discourse on AI governance in developing countries by focusing on the perspectives of legal professionals. It highlights the importance of regulatory approaches that are context-specific, inclusive, and capable of bridging the gap between global ethical principles and local realities. These insights offer practical guidance for policymakers, regulators, and scholars working to shape responsible AI governance in similar environments.

Low-Rank Adaptation (LoRA) enables parameter-efficient fine-tuning of Large Language Models (LLMs), and recent Mixture-of-Experts (MoE) extensions further enhance flexibility by dynamically combining multiple LoRA experts. However, existing MoE-augmented LoRA methods assume that experts operate independently, often leading to unstable routing, expert dominance. In this paper, we propose \textbf{TalkLoRA}, a communication-aware MoELoRA framework that relaxes this independence assumption by introducing expert-level communication prior to routing. TalkLoRA equips low-rank experts with a lightweight Talking Module that enables controlled information exchange across expert subspaces, producing a more robust global signal for routing. Theoretically, we show that expert communication smooths routing dynamics by mitigating perturbation amplification while strictly generalizing existing MoELoRA architectures. Empirically, TalkLoRA consistently outperforms vanilla LoRA and MoELoRA across diverse language understanding and generation tasks, achieving higher parameter efficiency and more balanced expert routing under comparable parameter budgets. These results highlight structured expert communication as a principled and effective enhancement for MoE-based parameter-efficient adaptation. Code is available at https://github.com/why0129/TalkLoRA.

Life Cycle Assessment (LCA) is increasingly used to quantify and regulate the environmental impacts of Information and Communication Technology (ICT) systems. Since direct biosphere measurements are complicated to perform, we claim that the environmental impact assessment of ICT relies heavily on models. In this paper, we first revisit the fundamentals of LCA: we emphasize that ICT LCAs effectively form systems of models, and we argue that such systems require an extra-high level of carefulness in construction, calibration, integration, and interpretation. We then document how this level of rigor is challenging to achieve with current practices. This is illustrated with emblematic examples of model misuse and an analysis of structural challenges related to database choice, scope mismatches, opaque aggregation, and model integration. From this analysis, we derive four key requirements for credible ICT LCA: explicit model lineage, clearly defined model scope, end-to-end traceability, and managed non-obsolescence. Finally, we propose a framework that operationalizes these requirements using explicit dependency graphs, an open and versioned LCA-oriented model repository, automatic enforcement of integrity constraints, and a well-defined model taxonomy.

In this paper, we analyze and improve the adversarial robustness of a convolutional neural network (CNN) that assists crystal-collimator alignment at CERN's Large Hadron Collider (LHC) by classifying a beam-loss monitor (BLM) time series during crystal rotation. We formalize a local robustness property for this classifier under an adversarial threat model based on real-world plausibility. Building on established parameterized input-transformation patterns used for transformation- and semantic-perturbation robustness, we instantiate a preprocessing-aware wrapper for our deployed time-series pipeline: we encode time-series normalization, padding constraints, and structured perturbations as a lightweight differentiable wrapper in front of the CNN, so that existing gradient-based robustness frameworks can operate on the deployed pipeline. For formal verification, data-dependent preprocessing such as per-window z-normalization introduces nonlinear operators that require verifier-specific abstractions. We therefore focus on attack-based robustness estimates and pipeline-checked validity by benchmarking robustness with the frameworks Foolbox and ART. Adversarial fine-tuning of the resulting CNN improves robust accuracy by up to 18.6 % without degrading clean accuracy. Finally, we extend robustness on time-series data beyond single windows to sequence-level robustness for sliding-window classification, introduce adversarial sequences as counterexamples to a temporal robustness requirement over full scans, and observe attack-induced misclassifications that persist across adjacent windows.

Mathematical models and numerical simulations offer a non-invasive way to explore cardiovascular phenomena, providing access to quantities that cannot be measured directly. In this study, we start with a one-dimensional multiscale blood flow model that describes the viscoelastic properties of arterial walls, and we focus on improving its practical applicability by addressing a major challenge: determining, in a reliable way, the viscoelastic parameters that control how arteries deform under pulsatile pressure. To achieve this, we employ Asymptotic-Preserving Neural Networks that embed the governing physical principles of the multiscale viscoelastic blood flow model within the learning procedure. This framework allows us to infer the viscoelastic parameters while simultaneously reconstructing the time-dependent evolution of the state variables of blood vessels. With this approach, pressure waveforms are estimated from readily accessible patient-specific data, i.e., cross-sectional area and velocity measurements from Doppler ultrasound, in vascular segments where direct pressure measurements are not available. Different numerical simulations, conducted in both synthetic and patient-specific scenarios, show the effectiveness of the proposed methodology.

Vision-Language Models (VLMs) have become essential for tasks such as image synthesis, captioning, and retrieval by aligning textual and visual information in a shared embedding space. Yet, this flexibility also makes them vulnerable to malicious prompts designed to produce unsafe content, raising critical safety concerns. Existing defenses either rely on blacklist filters, which are easily circumvented, or on heavy classifier-based systems, both of which are costly and fragile under embedding-level attacks. We address these challenges with two complementary components: Hyperbolic Prompt Espial (HyPE) and Hyperbolic Prompt Sanitization (HyPS). HyPE is a lightweight anomaly detector that leverages the structured geometry of hyperbolic space to model benign prompts and detect harmful ones as outliers. HyPS builds on this detection by applying explainable attribution methods to identify and selectively modify harmful words, neutralizing unsafe intent while preserving the original semantics of user prompts. Through extensive experiments across multiple datasets and adversarial scenarios, we prove that our framework consistently outperforms prior defenses in both detection accuracy and robustness. Together, HyPE and HyPS offer an efficient, interpretable, and resilient approach to safeguarding VLMs against malicious prompt misuse.

Autonomous AI agents powered by Large Language Models can reason, plan, and execute complex tasks, but their ability to autonomously retrieve information and run code introduces significant security risks. Existing approaches attempt to regulate agent behavior through training or prompting, which does not offer fundamental security guarantees. We present ClawLess, a security framework that enforces formally verified policies on AI agents under a worst-case threat model where the agent itself may be adversarial. ClawLess formalizes a fine-grained security model over system entities, trust scopes, and permissions to express dynamic policies that adapt to agents' runtime behavior. These policies are translated into concrete security rules and enforced through a user-space kernel augmented with BPF-based syscall interception. This approach bridges the formal security model with practical enforcement, ensuring security regardless of the agent's internal design.

We study mean estimation of a random vector $X$ in a distributed parameter-server-worker setup. Worker $i$ observes samples of $a_i^\top X$, where $a_i^\top$ is the $i$th row of a known sensing matrix $A$. The key challenges are adversarial measurements and asynchrony: a fixed subset of workers may transmit corrupted measurements, and workers are activated asynchronously--only one is active at any time. In our previous work, we proposed a two-timescale $\ell_1$-minimization algorithm and established asymptotic recovery under a null-space-property-like condition on $A$. In this work, we establish tight non-asymptotic convergence rates under the same null-space-property-like condition. We also identify relaxed conditions on $A$ under which exact recovery may fail but recovery of a projected component of $\mathbb{E}[X]$ remains possible. Overall, our results provide a unified finite-time characterization of robustness, identifiability, and statistical efficiency in distributed linear estimation with adversarial workers, with implications for network tomography and related distributed sensing problems.

Modern image compression methods are typically optimized for the rate--distortion--perception trade-off, whereas their robustness to bit-level corruption is rarely examined. We show that diffusion-based compressors built on the Reverse Channel Coding (RCC) paradigm are substantially more robust to bit flips than classical and learned codecs. We further introduce a more robust variant of Turbo-DDCM that significantly improves robustness while only minimally affecting the rate--distortion--perception trade-off. Our findings suggest that RCC-based compression can yield more resilient compressed representations, potentially reducing reliance on error-correcting codes in highly noisy environments.

Purpose: To develop and evaluate DosimeTron, an agentic AI system for automated patient-specific MC internal radiation dosimetry in PET/CT examinations. Materials and Methods: In this retrospective study, DosimeTron was evaluated on a publicly available PSMA-PET/CT dataset comprising 597 studies from 378 male patients acquired on three scanner models (18-F, n = 369; 68-Ga, n = 228). The system uses GPT-5.2 as its reasoning engine and 23 tools exposed via four Model Context Protocol servers, automating DICOM metadata extraction, image preprocessing, MC simulation, organ segmentation, and dosimetric reporting through natural-language interaction. Agentic performance was assessed using diverse prompt templates spanning single-turn instructions of varying specificity and multi-turn conversational exchanges, monitored via OpenTelemetry traces. Dosimetric accuracy was validated against OpenDose3D across 114 cases and 22 organs using Pearson's r, Lin's concordance correlation coefficient (CCC), and Bland-Altman analysis. Results: Across all prompt templates and all runs, no execution failures, pipeline errors, or hallucinated outputs were observed. Pearson's r ranged from 0.965 to 1.000 (median 0.997; all p < 0.001) and CCC from 0.963 to 1.000 (median 0.996). Mean absolute percentage difference was below 5% for 19 of 22 organs (median 2.5%). Total per-study processing time (SD) was 32.3 (6.0) minutes. Conclusion: DosimeTron autonomously executed complex dosimetry pipelines across diverse prompt configurations and achieved high dosimetric agreement with OpenDose3D at clinically acceptable processing times, demonstrating the feasibility of agentic AI for patient-specific Monte Carlo dosimetry in PET/CT.

Multimodal Sentiment Analysis (MSA) aims to infer human sentiment from textual, acoustic, and visual signals. In real-world scenarios, however, multimodal inputs are often compromised by dynamic noise or modality missingness. Existing methods typically treat these imperfections as discrete cases or assume fixed corruption ratios, which limits their adaptability to continuously varying reliability conditions. To address this, we first introduce a Continuous Reliability Spectrum to unify missingness and quality degradation into a single framework. Building on this, we propose QA-MoE, a Quality-Aware Mixture-of-Experts framework that quantifies modality reliability via self-supervised aleatoric uncertainty. This mechanism explicitly guides expert routing, enabling the model to suppress error propagation from unreliable signals while preserving task-relevant information. Extensive experiments indicate that QA-MoE achieves competitive or state-of-the-art performance across diverse degradation scenarios and exhibits a promising One-Checkpoint-for-All property in practice.

Accurate parameter selection is fundamental to gyrokinetic plasma simulations, yet current practices rely heavily on manual literature reviews, leading to inefficiencies and inconsistencies. We introduce Plasma GraphRAG, a novel framework that integrates Graph Retrieval-Augmented Generation (GraphRAG) with large language models (LLMs) for automated, physics-grounded parameter range identification. By constructing a domain-specific knowledge graph from curated plasma literature and enabling structured retrieval over graph-anchored entities and relations, Plasma GraphRAG enables LLMs to generate accurate, context-aware recommendations. Extensive evaluations across five metrics, comprehensiveness, diversity, grounding, hallucination, and empowerment, demonstrate that Plasma GraphRAG outperforms vanilla RAG by over $10\%$ in overall quality and reduces hallucination rates by up to $25\%$. {Beyond enhancing simulation reliability, Plasma GraphRAG offers a methodology for accelerating scientific discovery across complex, data-rich domains.