The Inference Report

How a vision-language model internally solves the task of describing an image is far from obvious. We find that the model develops a specific mechanism for this: a small set of attention heads in its language-model backbone, which we call gaze heads, whose attention tracks the image region the model is currently describing. We find them with a simple correlation score from a few forward passes, using comic strips as a controlled testbed where narrative order is laid out spatially. These gaze heads do not just track the image tokens being described: redirecting their attention to a chosen region forces the VLM to describe that region instead. A single attention-mask intervention on the top-100 gaze heads, fewer than 9% of all heads, steers the model's answer to any chosen comic panel at 83.1% accuracy, while the same intervention on random heads fails to redirect the answer, and intervening on all heads destroys generation. The same lever also extends to continuous control: switching the gaze target mid-generation makes the model wrap up its current panel description and move to the new one within a few tokens. Beyond comics, the same intervention redirects answers to chosen regions in natural COCO images. The mechanism further recurs across model sizes from 2B to 32B parameters and across other VLM architectures, although some frozen-encoder families show no comparable head set. More broadly, this shows that targeted edits identified through mechanistic analysis can serve as practical inference-time levers for steering multimodal model behavior, without any retraining. Our code, interactive demo, and datasets are available at https://gaze.baulab.info/

Building trustworthy medical multimodal large language models (MLLMs) is critical for reliable clinical decision support. Existing medical hallucination benchmarks mainly focus on data collection, but often ignore where hallucinations originate within the reasoning process. We find that hallucination sources vary across samples: errors may arise from visual misrecognition, incorrect medical knowledge recall, or flawed reasoning integration. To enable source-level hallucination diagnosis, we introduce ClinHallu, a benchmark for stage-wise hallucination diagnosis in medical MLLM reasoning. ClinHallu contains 7,031 validated instances, where each instance is augmented with a structured reasoning trace decomposed into Visual Recognition, Knowledge Recall, and Reasoning Integration. We also use stage-replacement interventions to measure how correcting specific stages affects the final answer. Beyond evaluation, we show that trace-supervised fine-tuning reduces stage-wise hallucinations. ClinHallu provides a fine-grained hallucination testbed for diagnosing and mitigating reasoning failures in medical MLLMs. The benchmark is publicly available at https://github.com/alibaba-damo-academy/ClinHallu.

Language Models (LMs) have shown remarkable potential as role-playing chatbots, delivering consistent, stylized interactions when given a specification of a character or user persona. However, applying these capabilities to real-world applications (e.g., ecosystems with numerous NPCs interacting simultaneously) exposes a critical inefficiency due to the excessive computational cost. In this paper, we question the necessity of dedicating a full, generalist model to a single persona, hypothesizing that a specific character identity relies on only a fraction of the model's total capacity. We observe that naively pruning LMs often severely degrades the role-playing performance for a specific persona; it does not distinguish between redundant knowledge and essential character traits. We propose Persona-Pruner, a framework that sculpts a lightweight role-playing model by isolating persona-specific sub-networks from a single description. Our experiments consistently show that Persona-Pruner preserves role-playing performance substantially more effectively than existing state-of-the-art LLM pruning techniques, reducing the performance drop from the dense model by up to 93.8% over the strongest baseline on RoleBench in LLM-as-a-judge score, while still maintaining general LLM capabilities. Code is available at https://github.com/jsu-kim/Persona-Pruner.

Large reasoning models typically follow a read-then-think paradigm: they observe the complete input, reason over a static context, and then produce the answer. Yet many real-world scenarios are inherently dynamic, such as audio and video stream, where information arrives as a continuous stream and models must reason, update, and respond under partial observations. Recent streaming reasoning methods allow models to think while reading, but they largely rely on supervised imitation of pre-constructed trajectories, which limits their flexibility. In this paper, we propose AdaSR, an adaptive streaming reasoning framework that enables models to reason during input streaming and perform final deliberation once the stream is complete, learning when to think, and how much computation to allocate across different stages. To optimize this hierarchical reasoning process, we introduce Hierarchical Relative Policy Optimization (HRPO), which decomposes policy optimization into streaming reasoning and deep reasoning phases, providing more fine-grained advantage assignment instead of uniformly distributing a single sequence-level advantage over all tokens. HRPO integrates format, accuracy, and adaptive thinking rewards to enforce valid reasoning protocols, preserve final task performance, and encourage latency-aware computation allocation. Experiments show that AdaSR achieves a better balance among reasoning accuracy, computational efficiency, and streaming latency compared with supervised fine-tuning baseline. We release our code at https://github.com/EIT-NLP/StreamingLLM/tree/main/AdaSR.

Cooperative multi-objective multi-agent reinforcement learning (MOMARL) models team decision making under multiple, potentially conflicting objectives. In this setting, conflicts arise not only across objectives but also across agents with different observations, roles, and contributions. We propose Preference Coordinated Multi-agent Policy Optimization (PCMA), which learns coordinated agent-specific preferences to enable complementary trade-offs among agents. Theoretically, we formulate cooperative MOMARL as a team-optimal game and show that, under suitable conditions, preference diversity can induce team improvement through a first-order improvement decomposition. Experiments on multiple cooperative MOMA environments and a practical traffic-control scenario show that PCMA improves both performance and trade-off coordination.

Reinforcement learning with verifiable rewards (RLVR) has successfully elicited the reasoning capabilities of large language models, motivating its extension to multimodal scenarios. Existing methods primarily focus on improving the visual coverage of reasoning traces and mitigating visual hallucinations, but underestimate the semantic inconsistency between the reasoning process and the final answer. In this paper, we delve into thinking-answer inconsistency in RLVR for large vision-language models (LVLMs), showing thorough analyses of rollouts collected throughout Group Relative Policy Optimization (GRPO) training process and post-RLVR evaluation outputs that this issue persists during training and remains present during inference. Motivated by the analysis, we propose Consistency-Oriented Reasoning Alignment (CORA), which introduces thinking-answer semantic consistency into RLVR through a lightweight plug-and-play consistency reward model, and further incorporates Hybrid Reward Advantage Splitting (HRAS) to stably coordinate task and consistency optimization. Extensive experiments across representative multimodal reasoning benchmarks and mainstream LVLMs show that CORA improves task performance while effectively mitigating thinking-answer inconsistency, leading to more faithful reasoning traces.

We study a \emph{max-risk} objective for active learning in a multi-group mean estimation $d$-armed bandits: a learner adaptively allocates a budget of $T$ samples across $d$ groups to minimize the worst-case uncertainty index $\max_{k\in[d]}σ_k^2/n_k$, where $σ_k$ is the standard deviation of the distribution of arm $d$, and $n_k$ is the number of times arm $d$ is sampled. We develop a local minimax framework and prove the first general lower bound for this objective, valid for any finite-variance hypothesis class. The bound separates difficulty into three orthogonal factors: a \emph{budget} term, a \emph{heteroscedasticity} index measuring how unevenly the uncertainty is spread across arms, and a model-dependent complexity measure, the \emph{Variance Local Curvature} ($\mathrm{VLC}$), which captures how much information a local change of variance creates inside the hypothesis class. For smooth classes, the $\mathrm{VLC}$ is a reparametrization of a variance--Fisher information, with closed-form values for common families. Benchmarking against the strongest available upper bound shows near-optimality up to logarithmic factors in broad regimes, and pinpoints a systematic gap in highly heterogeneous instances. Our proof introduces two key ingredients: a loss-induced $\ell_1$ geometry on the decision space, and a representation-based instance generator that reduces hard-instance construction to an explicit random matrix calculation.

AI systems coupled to proof assistants now generate formal mathematics at scale, and the gap between what a checker can verify and what a mathematician would value has become the binding constraint. We model the generation of valuable mathematics as nested language generation in the limit: a verifiable formal language $F$, accessed through a membership oracle (the proof checker), contains an unknown valuable language $H \in \mathcal{H}$ revealed only through an adversarial enumeration of a core $C \subseteq H$ of exact density $α$ (the literature). Every output is valuable ($\in H$), trivial ($\in F \setminus H$), or a hallucination ($\notin F$). We settle four questions. First, the verifier is not taste: the collections admitting generation with breadth are exactly those of the oracle-free model, characterized fiber-wise by Angluin's condition. Second, the verifier does buy sound coverage, covering all unseen valuable statements while asserting only valid ones: possible with it, impossible without it; it relocates unavoidable errors from false to trivial. Third, and centrally, a sharp dichotomy on the tight family: generators emitting finitely many trivia achieve optimal coverage $α/2$, while any infinite trivia allowance, even at vanishing rate, jumps the optimum to $1-α/2$ (both tight, for cores presented as the candidate intersection), and one generator attains both ends. The transition is in trivia count, not rate; the gap $1-α$ is the unrecorded mass. Fourth, both regimes instantiate in a compression model of mathematics. A perfect verifier cannot substitute for taste: the unbounded stream of correct-but-worthless statements is not an engineering accident but a provable necessity, since covering unrecorded valuable mathematics requires an infinite, but asymptotically negligible, stream of certified trivia.

Globally, cotton is a highly economically beneficial crop, as the textile industry heavily depends on it. So, the precise identification and detection of cotton leaf disease is crucial for economic stability. The development goal of "CottonLeafVision" is to accurately classify and detect cotton leaf disease. With this goal, we have evaluated multiple pretrained Deep Convolutional Neural Networks, including DenseNet201, InceptionV3, and VGG19 on a publicly available cotton leaf disease image dataset. This image dataset includes seven classes, six disease classes, and one healthy class, collected under various field conditions reflecting real-world challenges. Among these pretrained models, with DenseNet201, we have achieved the highest classification accuracy of 98%. To enhance the model reliability and interpretability, we have implemented different techniques and methods such as Gradient-weighted Class Activation Mapping (Grad-CAM), occlusion sensitivity analysis and adversarial training to increase the noise resistance of the model. Finally, we have developed a prototype in order to utilize the model's capabilities on real life agriculture. This paper shows the deep learning model's capabilities to classify the disease in real-life cotton disease management situations.

Real-time fire classification systems require models that are simultaneously accurate, computationally efficient, and deployable on resource-constrained hardware. This work proposes \textbf{HumP-KD}, a Hybrid Uncertainty-aware Multi-stage Progressive Knowledge Distillation framework for efficient fire classification. Two datasets, FlameVision and Dataset-II, containing 8,600 and 31,309 images, are used. Various CNN and transformer baselines are applied under standard preprocessing, online augmentation, Gaussian noise and motion blur robustness conditions. The proposed HumP-KD model distills knowledge from two frozen heterogeneous transformer teachers, Swin-Tiny and ViT-Base, along with their Meta-MLP ensemble, into a lightweight MobileViT-S student via three tightly integrated components. Hierarchical Progressive Knowledge Distillation employs a Hierarchical Feature Builder. It generates a fused spatial attention mask to guide distillation toward discriminative regions selectively. Multi-Stage Knowledge Distillation progressively activates three distillation stages across training. On Dataset-II, HumP-KD achieves a mean F1 score of $0.9876 \pm 0.0063$ across 10 independent trials, significantly outperforming the MobileViT-S baseline trained without distillation ($0.9537 \pm 0.0351$), with statistical significance confirmed by both independent t-test ($p = 0.0195$) and Wilcoxon signed-rank test ($W = 1$, $p = 0.0039$). The proposed method also demonstrates strong generalization across datasets and robustness under degraded visual conditions. The student model retains only 4.94M parameters and 19.01Mb model size, representing a $5.7\times$ parameter reduction over Swin-Tiny and a $17.5\times$ reduction over ViT-Base, while achieving 37.72 CPU FPS, making it suitable for real-time deployment.

Online inventory optimization (OIO) is online convex optimization with physical memory: inventory carryover makes the feasible action set depend on the past. A natural principle, used in stochastic inventory learning and recently in OIO under a single linear capacity constraint, is to maintain a hidden target chosen by an online learner and implement its projection onto the currently feasible order-up-to set. We prove that this simple principle is optimal for OIO on arbitrary bounded convex capacity sets. With online gradient descent as the base learner, the method improves the best known regret guarantee for OIO on general convex sets from inverse to inverse-square-root dependence on the common-demand probability, and we prove a matching lower bound. The same principle gives the first polylogarithmic regret guarantee for strongly convex losses and the first dynamic regret guarantee adapting to Euclidean path variation on general convex capacity sets. The analysis introduces a norm alignment principle: the right state variable is the distance from the hidden target to the feasible set, measured in the same norm as the projection. Under norm alignment, this distance evolves pathwise as a scalar queue, with target movement as arrival and common demand as service. This reduction to one-dimensional queue control resolves the state dependence and extends the guarantees to general convex capacity sets, beyond the reach of prior productwise approaches. Experiments on synthetic and real-world inventory data corroborate the theory.

A Detector Error Model (DEM) is a structured representation of error mechanisms in quantum circuits, which has gained popularity in quantum compilation pipelines for its ability to capture fault-tolerance at a circuit level. It lists error mechanisms as instructions targeting detectors and observables, specifying for each physical fault channel the probability that the fault fires, the detectors it triggers, and the observables it flips. In this paper, we develop an equational theory for DEMs, with its associated categorical semantics. We present a sound, terminating, confluent rewriting system for DEM terms, formulating it as a symmetric monoidal theory (a PROP) over the Giry monad. We prove that every DEM term has a unique normal form, which can be computed efficiently in quasilinear time $O(k|E|\log|E|)$, where $|E|$ is the number of instructions and $k$ bounds the size of a target set. This provides a complete set of invariants (via Tanner graphs) for structural DEM equivalence. We provide the first static decision procedure for DEM equivalence, with rigorous correctness guarantees. It is complete (decides full decoder-equivalence exactly) for non-adaptive quantum error correction (QEC) pipelines, and scales to a sound and applicable decision procedure for partially-adaptive circuits (lattice surgery, distributed QEC, ...) without suffering exponential overhead. We discuss its application to the verification and optimisation of quantum compilers.

LLM agents are increasingly built not as single model calls, but as scaffolded systems that combine reasoning, memory, reflection, action execution, and learning. While such scaffolds often improve performance, they are often embedded in tightly coupled pipelines, making it difficult to isolate component contributions, compare alternative designs, or understand how module interactions shape agent behavior. We introduce AgentSpec, a modular specification framework that represents embodied agents as typed compositions of reusable policy components with standardized interfaces. AgentSpec standardizes the interfaces among perception, memory, reasoning, reflection, action, and optional learning, enabling components to be swapped and recombined under controlled conditions. We instantiate this framework across DeliveryBench, ALFRED, MiniGrid, and RoboTHOR, and analyze reasoning, memory, reflection, and reinforcement-learning modules across model backbones. Our results show that agent performance is governed by scaffold compatibility and interaction effects rather than isolated module strength. In particular, structured multi-granularity memory improves long-horizon state tracking, reasoning and memory interact non-uniformly across environments, reflection trades off correction and cost, and RL-trained policies compose best when optimized with deployment-time scaffold structure. AgentSpec provides a controlled foundation for studying, comparing, and designing composable LLM agents. Our code, baselines and interactive playground are publicly available at https://agentspec-embodied.github.io.

We study whether the Compressed Computation (CC) toy model (Braun et al., 2025) is an instance of computation in superposition. The CC model appears to compute 100 ReLU functions with just 50 neurons, achieving a better loss than expected from only representing 50 ReLU functions. We show that the model mixes inputs via its noisy residual stream, corresponding to an unintended mixing matrix in the labels. Splitting the training objective into the ReLU term and the mixing term, we find that performance gains scale with the magnitude of the mixing matrix and vanish when the matrix is removed. The learned neuron directions concentrate in the subspace associated with the top 50 eigenvalues of the mixing matrix, suggesting that the mixing term governs the solution. Finally, a semi-non-negative matrix factorization (SNMF) baseline derived solely from the mixing matrix reproduces the qualitative loss profile and improves on prior baselines, though it does not match the trained model. These results suggest CC is not a suitable toy model of computation in superposition.

Large language models increasingly serve as execution engines for agentic systems, yet they still consume context through a sequential text interface. This creates a mismatch with modern structured agent workflows, in which independent branches explore subtasks, retrieve evidence, or generate candidate solutions before a final synthesis step. Existing systems typically merge these branches by concatenating their textual outputs, which discards the parallel structure and incurs redundant prefill computation. In this work, we introduce Parallel-Synthesis, a plug-and-play framework that enables a synthesizer to directly consume the KV caches produced by parallel worker agents. Parallel-Synthesis combines a cache mapper that calibrates independently generated branch caches with a fine-tuned synthesizer adapter that enables generation from this non-sequential cache interface. We train Parallel-Synthesis using data that exposes the synthesizer to parallel cache contexts, teaches aggregation across cached branches, and distills reasoning behavior from standard text-concatenation-based synthesis. Across nine downstream datasets spanning math, science QA, code generation, GAIA, and multi-agent database diagnosis, Parallel-Synthesis matches or outperforms text-based synthesis on seven datasets and remains close on the other two. It also reduces time-to-first-token by 2.5x-11x, suggesting that direct cache-based synthesis is a promising interface for more native and efficient synthesis over parallel agent branches.

Knowledge editing systems must update selected facts while preserving nearby but irrelevant behavior. This paper studies this problem in a memory-assisted setting where an edit memory is retrieved at inference time and a parameter-efficient adapter corrects the model's object preference. We argue that the central design question is not only how to write an edit, but also when to suppress it. We introduce \method{}, a route-specialized dual-adapter editor. A relevance router first decides whether a prompt should receive an edit memory. Routed prompts use an edit adapter trained to prefer the new object over the original object; unrouted non-direct prompts use a separate locality adapter trained to preserve or restore the original-object preference. We evaluate \method{} on three 1,000-case protocols, \cf{}, \zsre{}, and \mquake{}, under the same memory protocol and two 7B/8B base models. On Llama-3.1-8B-Instruct, \method{} obtains the best overall probability-preference accuracy on all three benchmarks: 0.8180 on \cf{}, 0.8946 on \zsre{}, and 0.9922 on \mquake{}. The same trend holds on Qwen3-8B. Router ablations show that the relevant memory boundary differs across datasets: a lexical neural router is safest on \cf{}, while BGE embedding routing is better on \zsre{} and \mquake{}. Component and module ablations show that the gain mainly comes from separating edit injection from off-route suppression rather than from simply increasing LoRA capacity.

Pretrained audio embeddings are standard in bioacoustics, yet little is known about which acoustic features these models encode, nor which are useful for a given task. This hinders transparency and limits extension to rare species or data-scarce domains. Here we reveal which speech-like features are encoded in bioacoustic representations. Using the 88~eGeMAPS features across six taxonomic groups, we apply linear and nonlinear regression probes to quantify which acoustic properties each model captures. Results confirm a ``no free lunch'' pattern: no single model captures the full feature space. A concatenated embedding achieves the highest performance, suggesting complementary acoustic space coverage across models. Loudness features are best encoded ($R^2 = 0.76$) while F0 is hardest to recover ($R^2 = 0.33$). By cross-referencing recoverability with per-species feature salience (NMI), we derive data-driven model selection guidance for bioacoustics.

Artificial Intelligence (AI) is increasingly used to automate a variety of real-world computer vision (CV) applications, such as autonomous vehicle control, facial recognition, and security cameras. Recent research has shown that acoustic vibration can induce real physical motion in cameras, interfering with their internal stabilization mechanisms. Because the motion falls outside the conditions the stabilization system was designed to handle, the system introduces artifacts into the frame, causing AI-based CV models to misclassify, miss targets, or hallucinate objects. Previous work used ultrasonic frequencies (>20 kHz) to perform short-range attacks, which limits them to short distances due to the attenuation exhibited by high frequencies. In this work, we investigate acoustic attacks using lower frequencies in the audible range (<20 kHz), and we further expand our analysis to include how various image and object features are affected by the attacks. Specifically, we performed physical experiments to demonstrate the viability of our attacks on an off-the-shelf object detection model (YOLO11) by resonating a commercially available camera with various frequencies. Based on our results, we provide insights into several factors that make an AI CV system more vulnerable to these attacks, which could help inform the development of future mitigation strategies.

Sequential or time-stamped interaction logs provide objective records of digital application usage, yet their granularity and noise often obscure meaningful insights into people's work. Such insights are essential for improving digital products in ways grounded in real-world user interactions. Prior research has applied deep learning models to cluster user actions into high-level activities, but these approaches are highly sensitive to noise and struggle to generalize across applications. To address this limitation, we introduce WorkflowView, a framework that uses large language models (LLMs) to abstract low-level action sequences into high-level activities. We establish the effectiveness and generality of our approach across three distinct, challenging sequential tasks and diverse domains: (a) zero-shot task description reconstruction from browser logs (achieving high semantic similarity, $μ_{sim} = 0.91$), (b) few-shot student dropout prediction using MOOC interaction logs (reaching weighted $F_1 = 0.90$ with only five few-shot examples), and (c) anonymized, privacy-preserving analysis of AI tool integration within document workflows in Microsoft Word. Our work demonstrates that LLM-based abstraction is a robust and efficient path forward for transforming low-level behavioral data into high-level, interpretable, and actionable insights. We also discuss practical considerations for deploying LLM-based inferences within logging infrastructures, including computational efficiency and user privacy.

The identification of optimal structures within vast arrays of interconnected data necessitates significant sampling- and computational effort. Learning and leveraging underlying signal dependencies can improve efficiency and predictive capabilities considerably, but the ubiquity of nonlinear statistical relations amplifies the complexity of such undertakings. In this paper, we develop novel generic and adaptive strategies equipped with routines for graph-based causal reward modeling, analytic reproducing kernel methods, and Taylor approximation of functional processes. We establish theoretical performance guarantees sublinear in time and linear in data volume over time. Our analyses cover robustness to a multitude of uncertainties arising from noise interference, gradual model convergence, and solution space mismatch. The framework's general appeal is substantiated by a minimalistic set of conditions or reliance on prior estimates, while various outlined modifications address specific or extended settings. To demonstrate practical effectiveness, we conduct numerical experiments using both benchmarked synthetic and real-world transportation datasets.

Robust machine learning and optimization rely on the uncertainty model choice. We investigate which uncertainty directions a model must cover when defined by a finite dictionary and a budget constraint. Selecting a subset forms an atomic uncertainty set with a closed form support function, yielding tractable robust programs for affine objectives. We propose a data driven selection rule based on a coverage objective over evaluation directions, including gradients, adversarial perturbations, or shifts observed on held out data. We prove this objective is monotone and submodular, supporting a greedy method with a $(1-1/e)$ approximation guarantee and a matching hardness barrier. We also provide a certificate bounding the loss from the selected subset and a radius calibration rule with out of sample control.

Transformer-based automatic speech recognition (ASR) models such as Whisper are highly accurate, but their predictions remain difficult to interpret. Existing explainable AI (XAI) methods often lack faithfulness and precise temporal grounding. We propose Listening with Entropy-guided Attention for Faithful explainability (LEAF-X), a model-intrinsic XAI framework for transformer-based ASR. LEAF-X combines entropy-guided attention weighting, multi-layer attention rollout, and optional causal ablations to identify low-entropy, high-impact heads and layers, producing sparse token-to-frame attributions. Unlike perturbation-based explainers or raw attention maps, LEAF-X exploits the internal structure of encoder-decoder and speech-augmented decoder-only models to generate explanations that better reflect model computation. Results show 32% improved faithfulness, 35-39% stronger locality/sparsity, and the most stable attributions, supporting more transparent and auditable ASR.

We study Online Convex Optimization (OCO) over a convex set $K\subseteq \mathbb R^d$, where in each round $t$ the learner selects $x_t\in K$ and then observes a convex loss $f_t:K\to[0,1]$, with the goal of minimizing regret to the best fixed decision in hindsight. We introduce a unified probing model that generalizes two recent lines of work: sublinear best-expert queries in the experts setting, and pairwise (comparison-based) feedback available every round in OCO. In our framework, the learner has a budget of $k\le T$ pairwise probes; on a probed round it may query two points and learn which one has smaller loss. Our main result shows that even a sublinear and noisy probe budget can provably improve worst-case regret in the full feedback OCO regime. With $k$ $δ$-noisy pairwise probes, we obtain: $ \text{Reg}_T \le O\left(\min\left\{\sqrt{dT\ln T},\; \frac{dT\ln T}{k|1-2δ|}\right\}\right) $, which is tight (up to logarithmic factors in $T$) across $T$, $k$ and $δ$. Specifically regarding the noise parameter $δ\in [0,1]$, the regret guarantee smoothly degrades as the oracle response approaches a coin flip, i.e., $δ$ is close to $\frac{1}{2}$. When applying the same techniques to a finite $K$ for the prediction with $d$ experts setting, the resulting rates are instead completely tight in all parameters, including $d$. Our analysis gives a streamlined treatment of pairwise probing in OCO by quantifying the benefit of probing via a variance reduction effect, combined with a second-order (variance-based) analysis of Continuous Exponential Weights.

Recent advances in speech generation have significantly improved the naturalness of synthetic speech, making spoofing detection increasingly challenging. A key limitation of current anti-spoofing systems is their limited robustness to unseen synthesis methods. In this work, we transform a self-supervised speech representation model into a Mixture-of-Experts (MoE) architecture to improve generalization. Feed-forward blocks in selected encoder layers are replaced by multiple expert networks controlled by a layer-wise gating mechanism, allowing experts to capture complementary acoustic patterns while preserving the representations learned during self-supervised pretraining. We further analyze the architectural choices affecting the performance of this MoE conversion and investigate the activation behavior of the experts. The proposed approach is evaluated on 14 spoofing datasets and reduces the macro EER from 5.46% to 4.81%, corresponding to 11.9% relative improvement over the baseline.

Control-function instrumental variable estimators need a first-stage residual, not merely a first-stage prediction. High-capacity first stages can interpolate treatment and leave too little residual information for the outcome equation. We study Adaptive Anisotropic Instrumental Heat Flow (A-IHF), a deterministic graph-diffusion residual extractor for flexible control functions. A-IHF treats treatment as a signal on a graph of first-stage features, uses pilot diffusion to detect large treatment jumps, attenuates conductance across those jumps, and computes the generated control with a sparse graph resolvent. Its observational selection rule uses only $(Z,X)$, combining graph generalized cross-validation, roughness, residualized-treatment relevance, and graph-admissibility filtering. The analysis decomposes error into structural leakage, residual attenuation, and residualized treatment variation, yielding finite-sample bounds, graph-admissibility rates under latent piecewise-smooth geometry, and finite-path selection calibration. Across 54 synthetic benchmark cells with tuned graph, kernel, tree, boosting, series, and neural control-function baselines, guarded observational A-IHF has the lowest average structural-response MSE; the A-IHF family beats the best non-A-IHF baseline in 32 cells. Performance is strongest when the graph captures piecewise-smooth first-stage structure.

Verifier-driven self-DPO is a common recipe for self-improving production visual-language models. In this setup, a frozen verifier scores candidate generations, the top- and bottom-scoring candidates form a preference example, and DPO updates the learner. The deployment-time assumption is monotone: a stronger verifier should yield a stronger student. We show that this assumption can fail because verifier quality is highly task-specific. On a four-rung open-source verifier ladder across MathVista, MMMU, and BLINK, the same verifiers that are above-threshold and improve a Qwen-3-VL-2B student on MathVista become sub-threshold on MMMU, where their task-rubric accuracy drops to 8% to 23%. In this regime, every verifier we tested silently regresses the student, producing drops of 3.4 to 10.9 percentage points below the frozen baseline while the DPO training loss continues to decrease. The regression replicates on a second student, Qwen-2.5-VL-3B. Moreover, within the failure regime, damage is confidence-inverted: the more accurate-but-still-wrong verifier causes larger regression than a near-random verifier, suggesting that progress-gated replay amplifies confidently wrong preference pairs. We give a compact mechanistic explanation via a variance theorem for progress-gated replay and its direction-mismatch failure mode. The deployment message is operational rather than purely diagnostic: before running any verifier-driven loop, teams should measure target-task rubric accuracy, rank verifiers by target-task rubric quality rather than parameter count, and treat diminishing returns in above-threshold regimes as a verifier-side compute budget cap.

The global use of artificial intelligence has increased interest in assessing the ability to generate culturally localized content, including stories. Cultural localization in stories often occurs through either templated localization -- the use of cultural markers (e.g., names, locations) in a generic narrative -- or holistic localization -- the variation of plots, values, and themes, in addition to cultural markers. We propose a method to measure the degree to which content was generated through templated localization. Specifically, we identify the lexical tokens that distinguish stories across nationalities and measure the similarity of the narratives that remain after removing them. In stories generated by five models on 125 topics for 193 nationalities, our method is able to detect that only a small subset (9-17%) of the vocabulary accounts for the variation across nationalities and that the narratives that remain after removing them contain repeated multi-word sequences, suggesting the presence of a shared culturally-agnostic narrative template. Finally, we characterize the cultural markers for their stereotypicality and offensiveness, finding that markers from 19 countries, mostly located in the Global South, are on average offensive.

Open diffusion language models are marketed as parallel, non-autoregressive decoders, yet the order in which a shipped checkpoint actually commits its tokens is almost never measured. We instrument DiffusionGemma 26B, a masked discrete-diffusion mixture-of-experts model built on Gemma 4, hooking its sampler's accept step to record which canvas positions commit, when, and at what confidence. Across a 686-prompt, six-regime probe suite we find that its decoding is neither parallel nor block-autoregressive: it follows a partial left-to-right commit bias whose apparent strength depends almost entirely on the granularity at which you look. Order is weak token by token and strengthens smoothly as the analysis is coarsened, so the model's "block size" turns out to be an artifact of the measuring ruler rather than the architecture. The model commits in large simultaneous batches, leaving much of the within-batch order genuinely undefined rather than merely unobserved. The behaviour is regime-dependent: structured JSON is committed in essentially arbitrary order, and a position's commit confidence tracks correctness on mathematical reasoning but carries no signal on factual recall. Commitment is aggressive, finishing in a short late burst well inside the step budget, while task accuracy matches the model's autoregressive Gemma-4 sibling. Beyond these findings, our central contribution is methodological: measuring decoding order honestly demands handling trailing-EOS padding, within-regime confounding, commit non-monotonicity, block-size sensitivity, and large commit-batch ties, each of which can otherwise manufacture a decoding-order result that is not really there.

Machine learning (ML) and Artificial Intelligence (AI) components are increasingly embedded in software products, yet undergraduate software engineering (SE) curricula rarely provide systematic preparation for building, testing, deploying, and maintaining AI/ML-based software systems. This paper aims to provide evidence-based guidance for integrating AI/MLrelevant content into core SE education. We compile and define a structured inventory of topics relevant to SE practice in AI/MLbased software, then map these topics against required courses in a set of representative SE curricula to identify coverage gaps. To assess educational priorities and feasibility, we survey SE instructors on topic importance and integration constraints. Based on the crosswalk between topic definitions, curriculum coverage, and instructor prioritization, we derive a guideline that recommends where and how high-priority topics can be embedded within existing SE courses.

Over time, GitHub has introduced different strategies for sharing reusable code artifacts. In addition to fork-based reuse, template repositories provide a distinct feature for generating new projects from scaffolding. Although this feature has been available since 2019, little is known about the domains it supports, its maintenance characteristics, or the practices that guide practitioners for effective template design. To address this gap, we conduct a large-scale empirical study of GitHub template repositories across the five most used programming languages. First, we mine and categorize templates to analyze the domains they serve, exploring the LLM-as-a-judge strategy. Next, we explore the reliability of templates by evaluating the associations between repository characteristics and activity, and quality-related issues (e.g., code smells, vulnerabilities, and security hotspots) through statistical analysis. Finally, we qualitatively analyze a representative subset of templates to derive practical guidelines and recurring pitfalls for template design and management. Our results show that Web Development is the predominant domain across ecosystems, while maintenance and quality issues vary by programming language. We further find that high-quality templates tend to adopt established software engineering practices, while providing comprehensive documentation and clear guidance for use. Overall, our findings offer empirical insights and actionable guidance to support practitioners in designing and adopting high-quality template repositories.

We show that the three movements of Beethoven's "Moonlight Sonata" (Op. 27 No. 2) instantiate three distinct machine learning architectures -- not by analogy, but by structural correspondence. Through computational analysis of the score (entropy, Jensen-Shannon divergence, dissonance, hand distributional overlap, self-similarity matrices, temporal memory decay, and contextual pitch embeddings), we establish four counterintuitive findings: (1) perceived musical "temperature" is governed by throughput, not distributional width; (2) the lightest movement carries the highest dissonance; (3) the movements implement streaming, recurrent, and periodic positional encoding memory architectures; and (4) the same pitch class acquires different contextual identities across movements, analogous to contextual vs.static embeddings in NLP -- and unsupervised clustering recovers the tonal structure without music-theoretic input. We construct a reverse sonification (decoding analytical features back into MIDI) and quantify the chirality of the encode-decode cycle: what distributions preserve and sequential ordering destroys. Prompted by a listener's observation that the decoded piece sounds like "mirror isomers that can't be superimposed," the chirality measurement reveals reconstruction loss increasing monotonically with n-gram order. Bootstrap baselines and subsample checks confirm all movements carry sequential information above noise, though raw values are confounded by sample size. Cross-domain comparison shows natural language has higher chirality than music, reflecting stronger sequential constraints.

Survival prediction plays a central role for healthcare providers and clinical researchers. Accurate risk stratification enables early intervention and improved patient management. Most existing deep survival models learn one common feature representation for all patients, which may hide important differences between patient subgroups. In contrast, a Mixture-of-Experts (MoE) framework allows different parts of the model to focus on different patient patterns, leading to more individualized representations. Therefore, in this work, we propose a mixture-of-experts enhanced adaptive deep clustering survival framework (AdaCSM) for modeling such heterogeneous survival patterns. We introduce a routing-based expert mechanism that enables conditional specialization within a parametric survival modeling framework. The proposed architecture allocates patients to specialized risk predictors dynamically while preserving the patient survival and subtype clustering objectives. We compare our method with state-of-the-art survival and deep clustering models on multiple real-world longitudinal clinical cohorts spanning diverse disease domains. The proposed method demonstrates improved predictive performance and leads to interpretable results in survival analysis.

Concurrent programming is a core component of Computer Science curricula, yet remains notoriously difficult for students to master due to its inherent complexity and the nondeterministic nature of concurrency bugs such as deadlocks and race conditions. In this work, we present ParaView, an educational tool designed to help students understand, debug, and correct concurrency issues in parallel programs written in C/C++. ParaView provides transparent execution recording and visualization to make parallel execution observable and comprehensible. We evaluated ParaView through a series of debugging and implementation tasks, with 17 students participating. Results showed a significant improvement in debugging and implementation successes compared to previous course iterations. A student survey confirmed that most participants found ParaView helpful. To further support learning outside the classroom, we explored using Large Language Models (LLMs) to analyze concurrency bugs and suggest fixes. While LLMs were highly effective in identifying bugs and explaining execution traces, the correctness of their bug fixes varied, especially for more complex synchronization patterns. Our findings suggest that recording-visualization tools like ParaView, complemented by artificial intelligence (AI), can improve teaching and learning of concurrent programming.

Wearable devices and smartphones generate rich behavioural time series that can support proactive health interventions, yet systematic comparisons of modern forecasting architectures for these data are lacking. In particular, it remains unclear how models generalise across populations, how different architectures respond to participant-level fine-tuning and how forecasting accuracy degrades across multi-day horizons. We benchmark six deep learning architectures, two zero-shot Foundation Models (FM) and statistical baselines on three public datasets encompassing over 800 participants, reporting per-feature metrics for step counts, screen time and sleep duration across 1-8 day horizons. We further conduct a per-feature personalisation study across all six architectures and assess FM transferability across dataset sizes and temporal granularities. Our key findings are: (i) no single architecture dominates, PatchTST leads among trained models while the three runners-up (TCN, MLP, Transformer) show no meaningful performance difference; (ii) the FM TimesFM matches or exceeds trained models zero-shot, especially in low-data regimes and (iii) participant-level fine-tuning reduces per-feature RMSE by 16-60\%, with sleep benefiting most and step counts least. These results provide practical guidance on architecture selection, FM applicability and personalisation strategies for mobile health forecasting. To the best of our knowledge, this is the first study to jointly evaluate modern deep learning, FMs and personalisation for multi-horizon behavioural forecasting from wearables.

This study presents a statistical and machine learning framework for characterizing a hydrogen-based multi-energy system (H-MES) using one year of high-resolution operational data. Statistical analysis revealed a binary operation driven by renewable surplus, with solar irradiance explaining 45.7% of rank-based variance in hydrogen production, a large effect by conventional standards. Only high-irradiance periods triggered meaningful electrolyzer engagement, while electricity demand exerted a weaker inverse suppression effect ($ε^2 = 0.126$). Multiple regression confirmed electrolyzer power as the dominant linear predictor, with a synergistic solar-wind interaction. Notably, Random Forest analysis ranked wind output first in predictive importance despite its weak bivariate correlation (r = 0.167), revealing non-linear dynamics invisible to parametric methods. A sequence model exploited strong 24-hour autocorrelation (r = 0.845) for operational forecasting, while a reinforcement learning agent optimized hydrogen revenue dispatch. The core contribution is demonstrating that statistical and machine learning approaches are complementary for H-MES modeling and control.

Online group chats are social spaces with local conversational norms that are rarely stated explicitly. The ability and willingness of LLM-based agents to recognize and adapt to these norms remains mostly unexplored. We introduce LoSoNA, a benchmark for local social norm adaptation in multi-party chat. Each scenario gives a subject model a curated group-chat transcript in which non-subject participants demonstrate a hidden local norm, followed by a final elicitor turn that forces a response revealing whether the subject has inferred that norm. We evaluate eight frontier and open-weight models under four prompting conditions that vary how explicitly the model is told to treat the prior conversation as evidence for how it should answer. Naive prompting remains limited for most models; explicit norm-aware prompting helps unevenly, with Gemini 3.1 Pro reaching $84.2\%$ and Claude Fable 5 reaching $81.6\%$, while several other models show small gains or regressions. LoSoNA contributes to recent calls for evaluating LLM social capabilities by testing whether models can infer local conversational norms from precedent and use them in a one-turn group-chat response.

Post-training INT8 (W8A8) quantization of diffusion transformers is widely deployed as a speed optimization, yet on consumer Ampere GPUs it is frequently slower than the FP8 and NF4 alternatives it is meant to beat. We trace this to a software artifact: the production "INT8" forward quantizes weights and activations only to immediately dequantize them back to bf16 and run a bf16 matrix multiply, never engaging the GPU's INT8 tensor cores, so the hardware's compute advantage is left entirely unrealized. We close this gap with a single fused Triton INT8 GEMM (int8xint8->int32 on Ampere tensor cores, with per-token x per-channel dequantization and bias folded into the epilogue, autotuned per GEMM shape) dropped into the Ideogram 4.0 diffusion transformer's linear layers in place of the dequantize-to-bf16 path. In the kernel, the int8xint8->int32 accumulation is bit-exact against torch._int_mm and the dequantized output matches the reference at cosine similarity 1.0 with no NaNs, running 2.8-4.2x faster than bf16 per GEMM. End to end it delivers a ~1.1x (~9-10%) speedup at 768px, and at 1024px it generates an image in 156.5 s on a single RTX 3090, faster than the single-card NF4 (164.5 s) and FP8 (172.9 s) baselines, at no measurable quality cost on these point estimates (PickScore/CLIPScore). INT8 thus goes from the slowest variant to the fastest, and 1024px becomes single-GPU feasible. The primary speed criterion (beat FP8, by ~9.5%) is comfortably met; the NF4 margin (~4.9%, single-run n=4) is within run-to-run variance we did not quantify and is best read as consistent with meeting the stretch target. We close with an honest deployment map: the win is specific to consumer Ampere, and on A100 and B200 the same kernel loses to those cards' fast native bf16/FP8 paths.

Transformer-based neural operators have shown remarkable performance for approximating solution operators of partial differential equations on complex geometries. However, existing approaches implicitly assume a fixed domain size, which limits their ability to generalize at inference. In this work, we investigate domain extension, namely zero-shot inference on spatial domains that are significantly larger than those encountered during training. We argue that this setting fundamentally requires spatial locality and translation equivariance. We propose to implement this locality via a decomposable bias in the attention logits computation, enabling finely controllable locality while remaining fully decomposable into query-key inner products and directly compatible with optimized attention kernels. Combined with rotary positional embeddings, it enables expressive embeddings with controllable spatial support without altering the transformer architecture. We empirically show that our approach substantially improves zero-shot generalization to larger domains across two PDE benchmarks and a 3D industrial atmospheric flow application. Our code and datasets are available at https://github.com/cerea-daml/domain-extension.

AI-assisted software development has moved from line-level autocomplete to agents that can plan changes, edit files, and submit pull requests with limited human supervision. Open-source software, however, evolves through a process designed for humans: contributor agreements, codes of conduct, and review norms all assume a legally accountable person who can attest to provenance and answer reviewer questions. Autonomous and semi-autonomous AI contributors strain those assumptions, and the 2025-2026 record of agent-driven incidents, AI-generated nuisance volume, and platform-level shutdowns shows that the gap is operationally consequential. Several open-source organisations have responded with contribution policies, but the result is fragmented, and its alignment with emerging AI governance frameworks (EU AI Act, NIST AI RMF with the UC Berkeley Agentic AI Profile, ISO/IEC 42001 and 23894) is unmapped at the contribution level. We compare policies across six organisations (SymPy, LLVM, matplotlib, OpenInfra, the Apache Software Foundation, and the Linux Foundation) using Most-Similar Systems Design with indicator-based coding and process tracing for SymPy and LLVM. From this we derive a six-dimensional taxonomy (disclosure, responsibility, human oversight, licensing, enforcement, maintainer workload), an ordinal Policy Maturity Score, and a mapping of documented agent incidents onto the dimensions each policy fails to govern. Aligning the dimensions with the regulatory frameworks above identifies overlapping gaps neither side currently closes, and we close by sketching the shape of a harmonised tiered framework and the empirical evaluation needed to calibrate it.

Clustering is widely used for exploratory analysis and scientific discovery, driving insights from market segmentation to biological data analysis, but its outputs can be difficult to interpret, audit, and reproduce as modern datasets become increasingly large and complex. Reliable use of clustering requires understanding which features drive the discovered structure, yet feature-level explanations for clustering remain scarce compared with methods in supervised learning. Furthermore, existing clustering feature importance scores are often tied to specific algorithms and data assumptions. To address these challenges, we propose Cluster LOCO (Leave-One-Covariate-Out), a family of model-agnostic feature importance scores for clustering. Cluster LOCO is built on feature occlusion and clustering generalizability, defined as whether cluster labels learned on one subset of the data can be accurately predicted on held-out samples. For any chosen clustering algorithm, Cluster LOCO quantifies a feature's importance by measuring how much its removal degrades generalizability. We first introduce Cluster LOCO-Split, which relies on data splitting, and then extend it to Cluster LOCO-MP, a minipatch ensemble-based version designed for large-scale data. Across synthetic simulations and an application to cell-type discovery in single-cell transcriptomics, we show that Cluster LOCO more reliably recovers informative features than existing clustering feature importance methods.

Large Audio-Language Models (LALMs) have shown strong performance on a wide range of audio understanding tasks, yet they still struggle with complex audio reasoning. A practical way to improve such capabilities is post-training, whose effectiveness critically depends on the quality and diversity of training data. However, existing audio-language datasets often contain substantial redundancy, where many samples are highly similar in acoustic content and thus provide overlapping supervisory signals. Such redundancy not only increases annotation cost, but also limits corpus diversity and reduces the effectiveness of post-training. To address this issue, we propose a redundancy-aware data construction pipeline for building reasoning-oriented supervision for LALMs. Specifically, we first perform acoustic similarity-based deduplication across raw audio datasets to improve corpus diversity. We then integrate existing audio captions and question-answer pairs into a unified multiple-choice format. Based on these unified annotations, we leverage Qwen3-30B to generate chain-of-thought (CoT) rationales for reasoning-oriented supervision. Based on this pipeline, we construct AudioDER, a reasoning-oriented post-training dataset containing approximately 191k samples spanning sound, speech, and music. Each sample consists of an audio clip, a multiple-choice question, four answer candidates, an audio caption, and a CoT rationale. Extensive experiments show that post-training on AudioDER consistently improves the performance of Qwen2-Audio-7B-Instruct on multiple audio reasoning benchmarks, including MMAU-mini, MMSU, and MMAR. We hope AudioDER can serve as a valuable resource for advancing audio reasoning research and the development of more capable LALMs.

LLM agent systems increasingly run as long-lived autonomous runtimes: scheduling jobs, calling tools, maintaining memory, and pushing results to humans. We present a longitudinal study of silent failures in one such system: a personal-assistant agent runtime in continuous production since March 2026, with roughly 40 scheduled jobs, 8 LLM providers, a tool-governance proxy, and a knowledge-base memory plane, defended by 4,286 unit tests and 827 governance checks. Over eight weeks we documented 22 incidents with full root-cause postmortems, in which one meta-pattern -- a failure whose error signal never reaches a human in actionable form -- manifested at least 28 times. We derive a five-class, mechanism-oriented taxonomy: (A) environment and platform quirks, (B) design-assumption mismatches, (C) error swallowing and dilution, (D) chained hallucination and fabrication, (E) operational omission and forensic blind spots. Class D is unique to LLM systems and the most dangerous: the system does not merely fail to report an error -- the LLM transforms it into fluent, plausible narrative delivered to the user. We term this fail-plausible: gray failure's differential observability escalated -- the observer is not just blind, it is convincingly lied to by the failure itself. Three findings: about 70% of silent failures were caught by human user-view observation, not tests or audits; a retrospective audit of 15 incidents found 0% ex-ante prevention but 87% regression blocking -- audits are regression engines, not prediction engines; incident latency (13 hours to 60 days) tracks failure mechanism, not code complexity -- the longest-lived failures lived in the seams between components, where no test runs. We describe the resulting defense framework and distill design principles for agent systems whose failures are loud, attributable, and boring. All postmortems and artifacts are public.

Generative dynamics models enable planning in challenging robotic systems, but safe deployment requires reliably detecting policy-induced out-of-distribution (OOD) transitions. Existing methods typically treat the learned dynamics as fixed and attach post hoc support surrogates. We show that these surrogates can fail when the dynamics are locally insensitive to critical action choices: unsupported control actions may produce latent predictions that resemble demonstrated transitions, suppressing OOD signals despite large true predictive errors. To address this, we introduce support-conditioned control-sensitivity regularization, which promotes sensitive local response to control input changes in learned dynamics in high-support training regions. This preserves control-induced variation while limiting unstable extrapolation due to weak empirical support. Experiments in vision-based obstacle avoidance, manipulation, and real-robot navigation show improved OOD detection and safer closed-loop planning.

Efficient route optimization play a vital role in ensuring both safety and punctuality in railway operations. It is very crucial particularly in heterogeneous multi-gauge railway networks with varying train speed, stopping pattern, infrastructure compatibility constraints increase coordination complexity. In single-track systems these challenges are further intensify due to all trains to share the same track and requires frequent track switching.Stochastic disruptions events including blocked tracks, blocked trains, engine failure and speed slowdowns introduces additional unpredictability in operations and deviate the timetable. However, existing studies predominantly focuses on high-level timetabling, omitting operational details such as track switching coordination. As a result leaving decision to human operators, increasing safety risks into railway operations. This study proposes a framework based on temporal planning for dynamic route optimization and disruption management in heterogeneous railway systems. The framework formulates railway operations as a temporal planning problem using PDDL 2.1 with explicitly modeling gauge compatibility constraints and diverse disruption scenarios. It generates conflict-free timestamped operational plans specifying both optimized schedules and executable action sequences. To evaluate the proposed framework, we developed a benchmark problem set with 200 instances using up to 1,000 track points and 120 trains. Two state-of-the-art temporal planners and a plan validator were employed to assessed the framework. The experimental results demonstrate that the framework effectively generates temporal operational plans for heterogeneous railway systems and handles multi-gauge constraints, disruptions, and reduces dependence on manual decision making.

Granting LLMs direct control over costly, irreversible scientific experiments leads to unsafe exploration and unstable performance, but discarding LLM creativity entirely sacrifices significant optimization potential. We introduce CARE (Controlling LLM-Generated Policies through Auditable Review of Evidence in Scientific Experimentation), an auditable controller for high-throughput experimentation (HTE) optimization that keeps a non-LLM incumbent optimizer as the default action path while using LLMs to revise challenger ranking policies. Before each outcome is revealed, a public-evidence intervention gate compares the challenger with the incumbent. It authorizes the challenger's selection only when the evidence available before selection supports the change, with the decision recorded in the audit log. CARE outperforms all other evaluated methods on Minerva/Olympus and ChemLex benchmarks, with final-best improving from 80.0 to 88.5 on Minerva/Olympus and from 83.9 to 92.1 on ChemLex, relative to the public incumbent. Our experiments indicate that LLM self-evolution is more reliable when it expands the proposal space under an auditable controller, rather than directly choosing experiments.

Identifying persuasive rhetorical cues is critical across domains, from detecting information manipulation and improving AI safety to advancing public health communication. We propose Persuasion Index (PI), a taxonomy of 15 dimensions grounded in persuasion theories from psychology and communication, and one transparent implementation using 55 sub-features built from lexicons and rule-based detectors. The taxonomy is modular: individual detectors can be replaced while preserving the theoretical structure. By evaluating PI on four public datasets varying in domain, style, and outcome measures, we show that PI provides a shared feature space for interpreting rhetorical patterns associated with persuasion-related outcomes. Linear models show that PI features carry meaningful predictive signal while remaining computationally lightweight. Dimension-level analyses reveal recurring associations between PI dimensions and persuasion outcomes across datasets, while also highlighting topic- and stance-specific variation. We release PI as an open-source package and web interface for principled and auditable analysis of human and AI-mediated communication.

When applying Group Relative Policy Optimization (GRPO) for GUI Grounding, rollouts are sampled from a single screenshot view; groups often become either all failures on difficult instances or all successes on easy ones, yielding no useful relative advantage. We propose VISTA (View-Consistent Self-Verified Training), a GRPO-based training framework that constructs each comparison group from multiple target-preserving views of the same GUI instance.Each view is generated by a crop that keeps the target element visible and remaps its box exactly, so model rollouts are compared across semantically equivalent but geometrically different inputs. To stabilize short coordinate generation without turning reinforcement learning into unconditional imitation, VISTA further adds a self-verified cross-view anchor: an oracle answer optimized with an advantage-weighted loss, excluded from the group baseline and activated only when the model has produced a maximum-reward rollout. Across five GUI-grounding benchmarks and multiple Qwen backbones, VISTA consistently improves grounding accuracy.On ScreenSpot-Pro, it raises Qwen3-VL 4B/8B/30B-A3B from 55.5/52.7/53.7 to 63.4/65.8/67.0. Robustness analyses further show higher worst-view accuracy and lower prediction flip rates.

Large language models (LLMs) are increasingly deployed as planners for autonomous agents in household environments. While existing benchmarks evaluate whether LLM-generated plans execute successfully, they overlook a critical type of failure: latent failures. Unlike immediate failures that trigger instant feedback at execution time and enable timely correction, latent failures do not immediately halt plan execution but silently compromise goal achievement. In severe cases, they cause irreversible harm. To address this gap, we introduce SIMMER, a benchmark for evaluating latent failures in LLM planning through a human-curated symbolic world model grounded in the kitchen domain. SIMMER defines a world model comprising 77 actions, 262 unique objects, and approximately 46,800 possible interactions that are semantically realistic, derived from real-world cooking scripts. It then leverages a state machine executor that validates plans against the world model and detects immediate precondition violations, latent hazards, and irreversible failures. Experiments across six LLMs show that even frontier models achieve at most 17% error-free plans. Moreover, up to 56% of plans contain latent failures, the majority of which lead to irreversible consequences. We further demonstrate that explicit state reasoning via counterfactual foresight simulation can reduce latent failures by up to 72% and irreversible cases by up to 75%, suggesting a promising direction for more robust LLM planners.

A central role of personal-agent memory is to turn stored information and prior interactions into future-oriented assistance. In daily use, useful cues come from what the agent observes and how the user interacts with the agent, and the agent must carry them forward from the current request to similar future tasks. Existing memory benchmarks usually test dialogue recall or task improvement in isolation, leaving the trajectory from streaming observations to later assistance largely untested. We introduce StreamMemBench, a streaming benchmark that constructs a two-step task sequence around each evidence anchor from EgoLife egocentric streams. The initial task tests evidence use, while the follow-up task tests whether feedback and interaction experience are reused. Four metrics diagnose evidence recall, initial evidence use, feedback incorporation, and follow-up reuse. Experiments with eight memory systems across two backbones show that current systems often fail to use observed evidence or turn feedback into reliable follow-up behavior, even when evidence is stored or feedback is incorporated locally. StreamMemBench is publicly available at https://github.com/landian60/StreamMemBench.

Emulators provide a cost-effective alternative to regional climate models (RCMs) by capturing their dynamical downscaling function. They link large-scale predictors simulated by global climate models (GCMs) to RCM-simulated high-resolution fields of the target variable, here precipitation. Machine learning methods, typically deep learning, are cheaper than running RCMs in computation time and energy. Among them, generative models are appealing because they can simulate ensembles of local high-resolution fields consistent with the predictors. This ensemble, which we call the uncertainty envelope, remains to be properly assessed for added value. Here, we make three contributions. First, we introduce ParamDiffusion, a new two-stage diffusion-based framework, and compare it with a state-of-the-art diffusion approach. Second, we expand standard validation through a comprehensive framework aligned with climate-science needs, examining specific precipitation events, including extremes. Third, within this framework, we assess the added value of diffusion approaches relative to deterministic methods. We intercompare four deep-learning models: a deterministic model designed to capture the precipitation tail; a parametric probabilistic model based on it; a recently proposed diffusion approach; and ParamDiffusion, which couples the parametric model with a diffusion model. Our results show that diffusion-based approaches reproduce climatological precipitation statistics with high skill, including distributional tails and spatially compounded extremes, while generating spatially detailed fields. However, none of the assessed models consistently accounts for the most extreme RCM-simulated events within its uncertainty envelope. Diffusion models are therefore promising for probabilistic RCM emulation, but progress is still required before they can reliably represent high-impact precipitation extremes.

Constitutive artificial neural networks (CANNs) provide interpretable material model discovery, but have so far been used in stress-supervised settings based on apparent stress-strain data from homogeneous tests. Because each test samples only a narrow loading path and provides homogenized rather than local stress information, robust discovery typically requires multiple loading modes to constrain the multidimensional response. This is challenging for soft biological tissues, where repeated testing, damage, and sample variability limit reliable information from a single specimen. Here, we combine CANNs with the stress-unsupervised full-field discovery framework EUCLID to identify sparse hyperelastic laws directly from displacement fields and reaction forces in one heterogeneity-inducing loading case. CANN-EUCLID minimizes equilibrium imbalance with sparsity-promoting regularization selecting compact active terms, without local stress measurements or a prescribed law. We evaluate the approach on isotropic and anisotropic benchmarks with prescribed ground-truth laws. When the ground truth is representable by the chosen CANN basis, our method recovers the correct terms with near-exact accuracy, including exponential terms with embedded parameters. When it is not contained in the basis, the method retains shared terms and approximates missing contributions using available basis functions. Generalization depends strongly on sampled deformation states: exponential strain-stiffening terms can be recovered accurately when sufficiently probed, but can produce large extrapolation errors when the stiffening regime lies outside the sampled domain. Forward FE validation simulations show that the discovered behavior accurately replicates the ground truth. These results establish stress-unsupervised CANN discovery as a promising framework for interpretable full-field constitutive model identification.

Sociable weaver nests function as complex ecological structures offering thermoregulatory microhabitats and sustaining diverse species; however, datasets used in prior studies lack fine-grained 3D structural detail. Producing usable and accurate 3D weaver nest data is challenging due to their irregular geometry and integration with complex host vegetation. We bridge this gap with an open-access, 1.4 TB multimodal drone dataset of 104 nest-bearing trees, comprising 27,945 RGB images, 111,780 multispectral images, approximately 781 million 3D points, and expert-annotated semantic segmentation labels. We benchmark semantic segmentation using KPConv, RandLA-Net, and Point Transformer V3, with PT-v3 achieving an mIoU of 86.35% on the test set. While the results demonstrate strong performance for transformer-based and point-wise methods, they also highlight architecture-dependent challenges, particularly for convolution-based approaches such as KPConv. By uniquely combining spectral, spatial, and structural information, the presented dataset advances 3D reconstruction, segmentation, and classification algorithms, enabling ecological applications from nest volume estimation to species conservation, and serves as a demanding benchmark that exposes architecture-dependent performance under extreme class imbalance.

Robotics manipulation research increasingly focuses on two-finger parallel grippers for their effectiveness, affordability, and ease of teleoperation. Grippers are nonetheless limited by their form factor, often requiring bimanual setups even for simple reorientation tasks. Anthropomorphic hands are a more natural platform for dexterous robot learning -- closer to the human hand, and capable of learning from human video -- yet they remain hard to use in learning research: even where open and accessible hand hardware exists, the software for control, simulation, teleoperation, and retargeting is scattered in one-off code bases, and largely disconnected from the robot-learning ecosystem. In this work, we introduce the \orca~learning stack, an open-source research stack for dexterity as a first-class robot learning domain. Our \orca~stack unifies low-level control, simulation, teleoperation from a range of consumer platforms, and hand retargeting, behind a single interface, and integrates natively with popular robot-learning frameworks such as \lerobot, so dexterous hand researchers can leverage the same data, training, and evaluation pipelines used for non-dexterous robot learning. We demonstrate a complete end-to-end workflow, collecting expert demonstrations of an in-hand reorientation task by teleoperation with a consumer-grade VR headset, training an autonomous policy with \lerobot, and evaluating the learned policy in a fully reproducible and observable setup. We open-source the entire stack as a shared, reproducible foundation for dexterous-manipulation research.

Non-Euclidean optimisation methods with matrix-valued updates, such as Muon and Scion, have recently shown strong empirical performance for training Transformer models, yet their theoretical advantages over Euclidean methods remain poorly understood. We address this gap in the heavy-tailed non-convex regime, where stochastic gradients have bounded $p$-th central moments, $p \in (1,2]$. We show that certain non-Euclidean methods achieve optimal sample complexity under stronger stationarity measures, while Euclidean methods incur additional dimension-dependent costs. As a consequence, for $m \times n$ matrices, Muon finds an $\varepsilon$-stationary point in nuclear norm within $\mathcal{O}\left(\min\{m, n\} \frac{Δ_1 L}{\varepsilon^2} \left(\frac σ\varepsilon \right)^{\frac p {p-1}}\right)$ samples, absorbing heavy-tailed noise without extra dimension dependence, unlike Euclidean methods. We further prove this sample complexity, including its dimension dependence, is optimal for all first-order methods under nuclear-norm stationarity. Experiments on large language models support our theory. Surprisingly, our results suggest that other Schatten geometries beyond the spectral geometry of Muon can perform competitively in certain settings.

Modern image classifiers widely adopt global average pooling (GAP) followed by a linear classification head. This linearity ensures that the image-level logits equal the average of logits obtained by applying the classification head pointwise to the feature grid prior to GAP. Consequently, standard classifiers may inherently retain spatial class evidence that remains recoverable even when the image-level prediction is incorrect. This structure naturally suggests a multiple-instance learning (MIL) interpretation, where an image is viewed as a bag of spatial instances. Within this formulation, we demonstrate that standard classifiers trained with a single label per image can still learn the intended classification task in multi-object scenes. We further exploit this property to decompose image-level logits into a prediction grid, providing a post-hoc diagnostic to extract spatial class evidence that GAP otherwise obscures. Our systematic evaluation reveals that off-the-shelf models consistently recover the ground-truth class within foreground regions. The MIL interpretation further suggests that common classifier failures reflect known limitations of mean aggregation.

Blockchain interoperability enables independent blockchain systems to communicate and exchange assets across heterogeneous networks. However, the lack of comprehensive security mechanisms remains a critical weakness -- one that attackers have already exploited to cause hundreds of millions of dollars in asset losses. This paper presents a systematic identification and classification of security threats facing interoperable blockchain systems, along with corresponding countermeasures for each. We organize threats into five categories: (1) core blockchain attacks, (2) network attacks, (3) interoperability-specific attacks, (4) social engineering, and (5) code vulnerabilities, with particular attention to smart contract weaknesses. For each identified threat, we analyze its attack surface and propose effective defensive strategies. The resulting taxonomy provides a structured foundation for designing and evaluating secure blockchain interoperability solutions.

Robots under autonomous operation may require decisions based on evidence that is no longer visible. We study \emph{delayed-evidence} tasks, where an early cue disappears before a later decision point, so visually similar observations can require different actions. In these settings, the current observation is not a sufficient state for control. We introduce TRAjectory-routed Causal Evidence (TRACE), a memory framework for visuomotor imitation policies. TRACE stores task-relevant visual and robot-state evidence, such as object identity, target choice, or route-dependent state, in a fixed-size latent memory that remains bounded over long episodes. Instead of indexing memory by raw time or manually provided task labels, TRACE uses \emph{path signatures}: compact, order-sensitive features of the executed robot-state trajectory. These signatures do not store the visual cue itself; rather, they provide trajectory-conditioned keys for writing and retrieving the evidence stored when the cue was visible. When the robot later reaches an ambiguous observation, the policy conditions on TRACE memory to recover the missing context and choose the correct branch. TRACE attaches through lightweight adapters to policies, without changing the policy backbone, action head, or imitation objective. Across real-world long-horizon manipulation tasks with visually ambiguous branch points, TRACE improves branch selection and task success over alternative baselines, including short-history and recurrent memory. Project page: https://jeong-zju.github.io/trace

Safe reinforcement learning (RL) aims to learn policies that optimize rewards while satisfying constraints. Predominant approaches rely on soft-constrained policy optimization, which has achieved empirical success but does not provide formal safety guarantees for the learned policy. In contrast, methods with strict guarantees typically rely on explicit certificate functions, whose construction requires the direct synthesis and verification of control-invariant sets, a process that scales poorly with state dimension and often yields overly conservative behavior. In this paper, we present the Provably Safe, yet Scalable RL (PS2-RL) framework, a novel two-phase architecture for learning provably safe policies in a scalable manner, designed to overcome the key bottlenecks of prior methods. Rather than explicitly computing invariant sets, PS2-RL leverages a learned backup policy to forward-integrate the system dynamics, generating an implicit control-invariant set online. In the first phase, the backup policy is trained with our proposed safe-arrival value function, which characterizes the optimal backup policy for invariant-set construction. In the second phase, an RL policy is trained end-to-end through a differentiable projection layer that strictly enforces the safety guarantees induced by the learned backup policy. By maximizing the volume of the implicit control-invariant set in the first phase, the resulting PS2 policy from the second phase is performant and scalable, while maintaining provable safety. Crucially, PS2-RL imposes no restrictions on the underlying RL algorithm and can be plugged into any existing training pipeline. We establish theoretical guarantees for the proposed framework and evaluate it on robotic control tasks with state dimensions up to 10, a regime in which prior provably safe RL methods struggle or become impractical.

Principal Component Analysis (PCA) preserves variance, not the information needed to detect rare catastrophic events. This paper proves the existence of a {\it Risk Shadow}: PCA can retain over 99.9999 percent of total variance while completely erasing all signal about rare, high-impact failures. When this happens, even the best possible classifier operating on the PCA representation reduces to a constant predictor. The root cause is a fundamental mismatch between variance maximization and tail risk awareness. To break the shadow, we introduce Expectile PCA (ExPCA) and Tail-Preserving PCA (TP-PCA), two methods that reweight the data covariance toward high-impact events. We prove theoretically that ExPCA strictly outperforms PCA in retaining rare-event information, and we validate our claims on synthetic data and a real-world credit card fraud detection benchmark. Our results call for a fundamental rethinking of variance-based dimensionality reduction in high-stakes decisions.

Large language models encode rich information in their hidden states. This work asks whether code correctness is legible in the hidden states of Qwen3-4B-Instruct-2507, before it generates and as it repairs a failed attempt, studied on 444 LiveCodeBench tasks. It reports two findings connected by a single confound-control tool: residualization. First, the correctness of the model's first-attempt code is linearly decodable from the prompt-final hidden state, with a leakage-free held-out AUC of 0.931 +/- 0.008 across 50 outer splits. After the linear effect of prompt length is removed from each hidden state dimension, the probe still reaches 0.911 +/- 0.010, well above a prompt-length baseline of 0.754 +/- 0.014. Second, on 236 cleaned cases where the model attempts to repair a failed first attempt, the hidden state shift from the failing attempt to its repair carries a statistically detectable contrastive direction, significant on both a magnitude and a split-half test against label-shuffled nulls. This direction does not survive a conditional residualization against repair-context covariates that differ between successful and failed repairs, marking it as a correlate of repair success driven by the repair context rather than an isolated repair-comprehension feature. The probe layer is selected by nested cross-validation, and the same residualization approach that upholds the pre-generation correctness result overturns the repair-direction interpretation. The contribution is as much methodological as empirical: a diagnostic honest enough to report a negative result alongside a positive one.

Real-time, full-duplex speech interaction is a key feature of next-generation spoken chatbots, allowing the model to listen and speak at the same time and to handle natural phenomena such as overlap, hesitation, and barge-in. Existing speech language models (SpeechLMs) such as LLaMA-Omni and GLM-4-Voice are still turn-based and rely on an external Voice Activity Detection (VAD) module to mark the end of the user's turn, which fundamentally limits their interactive ability. In this paper, we introduce BayLing-Duplex, a native full-duplex SpeechLM where a single autoregressive LLM decides when to listen, when to speak, and when to stop, with no auxiliary turn-taking module. The design adds only a few special tokens to the standard vocabulary, so it transfers across LLMs and reuses existing training and serving stacks with no architectural adaptation. Starting from the public GLM-4-Voice checkpoint and using only 400K full-duplex samples for fine-tuning followed by a lightweight DPO stage, BayLing-Duplex reaches 92% turn-taking success and 100% interruption success on InstructS2S-Eval, while improving the speech-response score from 2.17 to 3.39 over Moshi. BayLing-Duplex also matches or surpasses its turn-based counterpart on Llama Questions, Web Questions, and Alpaca-Eval, showing that simultaneous listen-and-speak modeling does not sacrifice response quality.

The removal of learned data from Machine Learning models through Machine Unlearning (MU) has been widely studied; however, there has yet to be an agreed-upon scheme for auditing MU. Existing work has shown that a dishonest model owner can falsify evidence to avoid executing MU, while curious auditors (and adversaries) can infer the privacy-sensitive properties of the model and its training data even with limited access. Yet auditing of MU under mutual distrust between the model owner and the auditor remains unexplored. We provide an information-theoretic proof for this scenario: for convex ML models, a generic audit scheme that relies solely on querying the model for \textit{behavioral} signals cannot identify insufficiently unlearned models without revealing membership information of the retained set. Therefore, auditing MU under the assumption of a dishonest model owner and an honest-but-curious auditor faces an inherent privacy-audit tradeoff. Our empirical results on convex models strongly supports this result, while further experiments demonstrate that this privacy-audit tension persists in non-convex models. Our results call for a more careful consideration of the privacy-audit tension under a realistic auditor threat model, and serve as a foundation for more scrutiny of designs of privacy-preserving audit schemes for the MU pipeline. We also release our code implementation at https://github.com/LiouTang/Behavioral-Unlearn-Audit.

LLM-based guardrails have emerged as a highly effective defense against prompt injection and jailbreak attacks in autonomous agents. However, we reveal that the very reasoning and task-following capabilities enabling this protection introduce a novel vulnerability: attackers can inject crafted data to trap the guardrail in extended reasoning loops, effectuating a systematic denial-of-service (DoS) attack. To systematically expose this threat, we design a beam-search optimization framework that crafts natural-language payloads to maximize guardrail reasoning length, utilizing an LLM proposer guided by a strategy bank. Based on the observation of guardrail's schema-following nature, we also provide another attack framework driven by mechanism-aware structural mutations with less computational load. The attack efficacy is systematically evaluated in two parts. First, in standalone evaluations, the attack generalizes across diverse guardrail architectures, safety templates, and agent benchmarks. Payloads optimized on a single open-source surrogate successfully transfer to eight leading model backbones (e.g., Claude, GPT, Gemini, DeepSeek, and Qwen), achieving a 13--63$\times$ token amplification. Second, in end-to-end real-world agent deployments (web, desktop, code, and multi-agent systems), the attack reveals up to a 148$\times$ latency amplification. We show that a single poisoned document can saturate shared guardrail infrastructures, effectively starving co-located agents and paralyzing the entire system. By uncovering this availability flaw, our work underscores the urgent need to develop cost-bounded, reasoning-robust guardrails.

AI evaluations are widely used for testing and understanding progress. However, the diverse evaluators bring with them inconsistencies that challenge analysis and comparison. First, results are saved in incompatible formats, scattered across leaderboards, papers, blog posts, evaluation harness logs, and custom repositories. Second, results are created by different evaluation frameworks, which produce divergent scores for nominally identical evaluations and record metadata inconsistently, hindering comparison, cross-community evaluation science, cost reduction, and reuse. We introduce Every Eval Ever, the first shared schema and community-crowdsourced repository for AI evaluation results. The schema standardizes how evaluations are represented in a unified, single JSON document. It is source-agnostic by design, ingesting results from evaluation harnesses and papers alike, and optionally stores per-instance outputs for fine-grained analysis. We contribute: (i) a community-governed metadata schema with a companion instance-level schema, the first standardization effort of its kind; (ii) automatic converters from popular formats, evaluation harnesses, and leaderboards to the unified schema; and (iii) a crowdsourced community database hosted on Hugging Face, currently spanning to date 22,235 models, 2,273 unique benchmarks, and 31 evaluation formats.

Internet of Medical Things (IoMT) devices operate under strict resource constraints while handling highly sensitive health data, making security and privacy critical concerns. Federated learning (FL) further complicates this landscape, as model updates exchanged during training may unintentionally expose private medical information. Emerging quantum computing capabilities threaten the long-term viability of conventional lightweight cryptographic mechanisms, motivating the integration of Post-Quantum Cryptography (PQC) into IoMT systems. This article discusses key enabling technologies for quantum-resilient IoMT, including post-quantum key establishment, lightweight encryption, and edge-native orchestration. We propose a scalable Kubernetes-based framework that integrates PQC into FL-enabled IoMT environments and validate it on a Raspberry Pi testbed. Results demonstrate that distributed cryptographic processing significantly reduces latency compared to sequential designs while maintaining feasible resource overhead. The primary contribution of this work lies in the design and validation of a secure orchestration and communication framework for FL-enabled IoMT systems. We conclude by outlining future directions toward energy-aware architectures, intelligent security optimization, and resilient next-generation Intelligent Internet of Medical Things (IIoMT) ecosystems.

The recent successes of neural networks producing human-like language have caused significant stir in cognitive science, with many researchers arguing that classical puzzles about human cognition and challenges to artificial intelligence are being solved by neural networks. A notable case is the argument from systematicity due to Jerry Fodor and Zenon Pylyshyn, argues that humans display systematic biconditional dependencies. For example, someone can understand the sentence "John saw Mary" just in case that they understand the sentence "Mary saw John." Symbolic systems explain this systematicity of language and thought, while neural networks offer no immediate explanation. Several recent articles argue that this challenge has now been met by neural networks. In particular, Brenden Lake and Marco Baroni argue that their meta-learning for compositionality protocol matches and perhaps explains human systematicity. We demonstrate that these conclusions are premature. Among other results, we found that their model struggles to learn rules that are even slightly out of distribution compared to their training data. Furthermore, the model behaves unsystematically even on many within-distribution problems. We conclude that Fodor and Pylyshyn's challenge to neural networks remains unmet.

Macrocyclic peptides are promising therapeutic candidates for intracellular targets, but their design requires simultaneous control over non-natural monomer chemistry, ring topology, membrane permeability, and target binding. Existing SMILES- or HELM-string generative models either operate in long atom-level sequence spaces or treat monomers as symbolic tokens with limited chemical grounding. We introduce PepALD, an Autoregressive Latent Diffusion (ALD) foundation model for \textit{de novo} macrocyclic peptide generation. The model represents HELM monomers with structured chemical embeddings, generates each residue through context-conditioned diffusion in chemically informed latent space, predicts R-group-aware ring closures during autoregressive generation, and aligns the denoiser to affinity rewards using winner-protected diffusion-adapted preference optimization. In silico experiments demonstrate PepALD's generation quality and reward-optimization performance against representative peptide generation baselines.

Fine-tuning vision-language models to emit dense coordinate lists improves visual grounding but also changes how models serialize, repeat, and terminate structured outputs. We study this behavior as a generation and control surface. In Gemma 4 12B, high-capacity q/k/v/o LoRA raises class-aware F1@0.3 from 0.007 to 0.448 while inducing repeated-tail pressure (duplicate rate 0.080, max repeat 23). A q/v rank sweep keeps max repeat at 21-22 across ranks 4-64, showing capacity persistence. The target signal is separable: object-level repeat-stop removes exact repeated records (duplicate rate 0.000, max repeat 1) while preserving F1 (0.494 to 0.490) and stricter F1@0.5 (0.381 to 0.385). Structure-axis probes localize the effect to bbox-coordinate object lists; dense non-bbox and spatial/count JSON remain repeat-clean, including under high-capacity adapters. Qwen3-VL-8B reproduces a clean controlled endpoint (F1@0.3 0.318, duplicate rate 0.000), and COCO 2017 reproduces acquisition plus duplicate pressure. Dense coordinate-list adaptation therefore creates a structure-bound, cross-family interference surface that can be measured and controlled.

Understanding how a prediction model will perform in a new environment before deployment is essential to preventing harm when algorithms inform decision-making. Two common sources of model performance degradation are (i) covariate shift, where the target covariate distribution differs from the source, and (ii) selective labels, where the observability of outcomes depends on historical decisions. We study pre-deployment model evaluation under the joint presence of covariate shift and labeling of outcomes selectively based on observed features. In particular, we present a double machine learning procedure for estimating the target risk of an arbitrary black-box prediction model under a general loss function. We show identification of this estimand under standard assumptions and derive a bias-corrected estimator based on the influence function of the target risk. Finally, we evaluate our estimator through experiments using the eICU electronic health records database, showing that it tracks the true target risk more accurately than methods that address either selective labels or covariate shift alone, as well as baselines that combine standard plug-in approaches.

Large Language Models (LLMs) are undergoing a fundamental transformation from conversational generators into integrated AI systems capable of reasoning, action, memory, and self-improvement. We conceptualize this transition as a shift from Chatbot to Digital Colleague: from conversational answers to persistent work. We organize this transition along two tightly coupled dimensions. First, at the cognitive core level, LLMs are advancing from Chatbot-era "fast thinking" systems driven by next-token prediction toward Thinking LLMs that leverage inference-time computation, Chain-of-Thought reasoning, reflection, process supervision, and reinforcement learning to support more deliberate and reliable cognition. Second, at the tool-augmented task execution level, LLMs are progressing from tool-calling Agents that invoke external resources in an ad hoc manner toward OpenClaw-style workstation systems (OpenClaw) equipped with persistent Workspaces, skills, verification loops, and governance. The "Workspace + Skill" paradigm makes episodic tool use colleague-like via state persistence, reusable procedures, task closure, and experience reuse. We examine data construction shifts from instruction-response pairs to State-Action-Observation trajectories and evaluation from static benchmarks to sandboxed, auditable, self-evolving AI ecosystems.

Predicting the Kohn-Sham Hamiltonian with machine learning can accelerate density functional theory while retaining access to molecular orbitals, energy levels, and electronic-structure observables that energy-only surrogates cannot resolve. Yet element-wise agreement with the converged Hamiltonian, an implicit fixed point of the self-consistent field iteration, does not determine the occupied subspace that governs orbital energies and densities. Here we present HamEvo, a neural operator that learns the single-step self-consistent update and returns the converged Hamiltonian as its fixed point. HamEvo is pre-trained on intermediate self-consistent trajectories and calibrated at equilibrium with density-matrix supervision. Across benchmarks from MD17 to drug-like QMugs, HamEvo lowers Hamiltonian errors by 35-49% over direct-regression and deep-equilibrium baselines, and predicts QMugs HOMO and LUMO energies with mean absolute errors of 0.036 and 0.053 eV, near the 1 kcal/mol chemical-accuracy scale. Few-shot fine-tuning with only 20 reference conformations extends HamEvo to molecules of up to 122 atoms, well beyond the size range covered by pre-training. With thermal molecular-dynamics sampling, HamEvo captures temperature-dependent HOMO-LUMO gap renormalization beyond the harmonic approximation. Inference is up to 242 times faster than conventional DFT.

We present a recipe-controlled decoder audit (RCDA) for structural transductive knowledge-graph completion (KGC). The audit asks a simple reporting question: before attributing gains to an encoder or training recipe, what changes when the decoder is swapped under the same recipe? Using ComplEx and DistMult as the primary controlled pair, with targeted RotatE/TransE spot-checks, we evaluate seven benchmarks. On five standard KGs, ComplEx-vs-DistMult differences are modest but consistent under our recipe (+0.005 to +0.012 MRR), whereas CompGCN-style encoder effects vary more by dataset. On small KGs, decoder effects become the main diagnostic: Kinship shows a stable ComplEx advantage of +0.143 MRR (6 seeds), while UMLS favours ComplEx by +0.022 MRR in a clean 6-seed server rerun but reverses in an earlier provenance variant. We therefore treat small-KG decoder choice as recipe- and provenance-sensitive rather than as a fixed dataset winner. We further show that decoder choice interacts with encoder depth on WN18RR, and that under our recipe L=0 ComplEx on YAGO3-10 reaches 0.6971 +/- 0.0048 MRR at d=128. The result is a compact audit protocol: report matched decoder rows, log small-KG provenance, and sweep decoder x depth before making encoder-level claims.

Recent finite-time analyses of nonlinear two-time-scale stochastic approximation show that under contractive assumptions the slow iterate $Y_k$ with stepsizes $β_k=Θ(k^{-1})$ and $α_k=Θ(k^{-a})$, $a\in(1/2,1)$, generally satisfies a mean-square rate of order $k^{-a}$; decoupled $k^{-1}$ rates require strong local linearity. We identify a sharp regularity-dependent boundary. In a rate-determining normal form where the slow drift contains a locally linear leakage and a nonlinear remainder of order $1+ρ$ ($ρ\in[0,1]$), the uncorrected recursion satisfies \[ \mathbb{E}\|Y_k\|^2 \le C\bigl(k^{-1}+k^{-a(1+ρ)}\bigr), \] and a matching scalar Gaussian lower bound shows that the slower term is unavoidable without modifying the update. Thus the decoupled $k^{-1}$ rate is guaranteed for the uncorrected recursion exactly when $a(1+ρ)\ge 1$. This lower bound concerns only the naive update; it is not an information-theoretic obstruction. We demonstrate this by equipping the normal-form recursion with an auxiliary online bias estimator \[ M_{k+1}=M_k+γ_k(R(X_k)-M_k),\qquad β_k\llγ_k\llα_k, \] and subtracting $M_k$ from the slow update. Under the same stability, moment, and remainder assumptions, the corrected recursion achieves $\mathbb{E}\|\widetilde Y_k\|^2=O(k^{-1})$ for every $ρ\in[0,1]$, including regimes where the uncorrected update provably suffers the slower rate. Finally, we prove localized transfer theorems that extend the phase-transition mechanism to general nonlinear TTSA in fast-manifold coordinates. The proofs are non-asymptotic and rely on two Abel-transform cancellations: one for the locally linear fast-error leakage, and one for the tracked nonlinear bias.

A growing line of work equips large language model (LLM) agents with graph neural networks (GNNs) as callable tools, assuming the agent exercises judgment over when and how much to rely on such a tool. We test this directly. We expose a frozen GNN to a ReAct-style LLM agent as an explicit tool and measure, on node classification over a text-attributed graph (ogbn-arxiv, replicated on WikiCS), whether the agent uses the tool or merely obeys it. We find the agent does not exercise judgment: its predictions agree with the raw GNN's 97.6-99.2% of the time (5 seeds), collapsing into a GNN parrot that adopts the tool's output wholesale and bypasses its own reasoning. Sweeping backbone capability (Qwen2.5 0.5B-7B), the deference is not a weak-model artifact: among models able to invoke the tool, agreement rises with capability (0.60 to 0.98 from 1.5B to 7B). Crucially, the cost of deference does not shrink as capability grows and grows where alternatives emerge: a per-node oracle over the available actions beats the parrot by 0.09-0.18 at 3B and 0.12-0.22 at 7B, roughly doubling at high homophily, because the parrot is pinned to the frozen GNN while the agent's alternatives improve; at 7B a simple neighbour-label tool overtakes the GNN at high homophily (0.81 vs 0.71) yet the agent still defers. A simple selective-invocation gate recovers about half of that high-homophily gap (0.71 to 0.83) but yields no net global gain, and held-out estimates bound the best achievable gate over standard test-time features to at most a third of the oracle headroom: reliable selective invocation looks limited by available information, not merely router design. Our results are a cautionary measurement: evaluations of agent+tool systems cannot assume the agent adds judgment on top of the tool, and selective invocation must be designed in rather than expected to emerge from scale.

Large language model (LLM) reasoning is ephemeral: chains of thought vanish with the context window, pruned search branches leave no record, and memory buffers cannot be diffed, merged, or audited. Every other complex software process (code, infrastructure, data, experiments) is version-controlled; reasoning is not. We introduce GitOfThoughts, which stores an agent's reasoning tree as a git repository: every scored thought is a commit, scores are notes, outcomes are tags, and retrieval is "git log" over the agent's own history. This makes reasoning replayable, auditable, and mergeable across agents at near-zero engineering cost. We then ask the harder question: does memory, in any substrate, actually improve accuracy? Across five substrates (none, markdown, vector, graph, git), two benchmarks, two model scales, and pre-registered replications, the answer for novel problems is no. No memory format reliably helps, and a promising early result collapsed under its own pre-registered replication. Memory pays only above what we call the copyability threshold: when the retrieved case is a near-duplicate of the current problem (similarity >~ 0.8), accuracy jumps sharply; below it, nothing. The gain is answer retrieval, not method transfer: a 4.5x larger model doubles the near-duplicate payoff yet still cannot extract a transferable method from a worked example. The only general lever we find is test-time sampling. The case for git-as-substrate is therefore auditability, provenance, and mergeability at accuracy parity. We document a retracted result and a refuted hypothesis to model the evaluation standard we hold ourselves to.

This paper investigates the fragility of post-hoc explanation methods in audio deepfake detection. While previous work on explanation manipulation focused on images using standard $L_p$ metrics, we introduce a psychoacoustic framework that optimizes inaudible perturbations to decouple model attributions from final classifications. We evaluate this vulnerability across state-of-the-art architectures under strict prediction-preserving constraints. By evaluating the manipulation cost through domain-specific perceptual audio quality metrics alongside explanation alignment criteria, our framework demonstrates that an adversary can systematically distort automated explanation heatmaps while preserving the predicted deepfake label. Full code available at: https://github.com/cncPomper/Audio-XAI

Neurosymbolic (NeSy) models integrate neural networks and symbolic reasoning for robust and interpretable AI. State-of-the-art NeSy models require that the symbolic component is expressed in a differentiable way, often complicating the use of approximate inference. We propose EM-NeSy which casts probabilistic NeSy learning as an instance of the Expectation-Maximization (EM) algorithm. In the expectation step, we compute the posterior over the neurally predicted symbols conditioned on the label via probabilistic inference. In the maximization step, we update the neural parameters based on this posterior using gradient descent only through the neural component. This formulation unlocks the full potential of the EM algorithm for NeSy learning. It allows NeSy to extend naturally to approximate reasoning without any additional modifications or differentiability requirements of the symbolic component. Furthermore, it recovers the standard end-to-end gradient-based NeSy setting under exact inference. Our experimental results demonstrate the scalability and computational efficiency of EM-NeSy.

Transformer-based clinical language models are increasingly integrated into high-stakes clinical decision support pipelines, yet the computational mechanisms through which demographic associations encoded in medical documentation propagate into model probability distributions remain empirically underspecified. We present a systematic computational audit of representational bias in ClinicalBERT (Alsentzer et al., 2019), a BERT-based model pretrained on MIMIC-III discharge summaries, employing two complementary probing methodologies: Log Probability Bias Analysis (LPBA), which quantifies demographic descriptor-induced shifts in masked token probability distributions across behavioral and evaluative semantic categories, and Masked Language Model-based analysis (MLM), which probes internal representational structure for demographic agency attribution encoding across 98 real clinical sentence templates and eight intersectional race-gender combinations. Corpus frequency analysis operationalizes the distinction between statistical disparity and bias amplification by benchmarking model outputs against empirical term frequencies in the MIMIC-III training corpus. Of 32 statistically significant findings, 65.6% contradict observed corpus distributions, rising to 80% for Black patients and 87.5% for agency attribution under MLM probing, providing direct empirical evidence that representational bias in ClinicalBERT operates predominantly through model-internal amplification rather than training data inheritance. Keywords: natural language processing, clinical documentation, algorithmic auditing, representational bias, health equity 1

Modern Automatic Speech Recognition (ASR) systems have made remarkable progress on standard benchmarks, yet performance gaps have emerged under real-world distribution shifts, caused by recording conditions, accents, speech impairments, and noise. Existing datasets and benchmarks typically isolate these factors, which overlooks their co-occurrence in real-world applications. In this paper, we argue that model robustness can be treated as a dynamic capability that continually develops, and we introduce MoDiCoL, a Modular Diagnostic Continual Learning dataset designed for controlled analysis of linguistic content, speaker characteristics, and acoustic environments. Furthermore, we propose a real-world-inspired continual learning curriculum to simulate incremental updates and study how robustness is acquired, transferred, and forgotten. We evaluate three continual learning strategies and provide detailed insights into robustness under evolving conditions.

Existing multi-agent software development systems have proposed many forms of agent collaboration, including role-based collaboration and automated code review. However, many systems assume a common runtime, a central conversation server, or the same API family. Under these assumptions, LLM agents from different vendors cannot easily exchange messages directly from their own execution environments while dividing development and review work on a shared codebase. This paper presents tap, a file-based collaboration protocol that allows Claude (Anthropic) and Codex (OpenAI) to collaborate on one codebase without shared memory or an identical runtime. The core of tap is a file-first design that preserves markdown files with metadata as original messages, combines a file inspection path (file communication, Tier 1) with real-time notification paths for Claude and Codex (real-time communication, Tier 2), and isolates work through separate git worktrees. Even if real-time notification fails or a receiver restarts, the message file remains available and the same content can be inspected again. In a 27-day, 37-generation self-applied operation where tap was used to develop and review itself, we collected 209 tap-related pull requests and 717 operational artifacts. An analysis of 375 review artifacts showed that the share of reviews recording at least one defect or requested change was 69.8% for heterogeneous model pairs and 53.1% for homogeneous model pairs. These results show that tap, which combines file-based message preservation with real-time notification, operates in a real production repository, and that combining heterogeneous models and execution environments can broaden review perspectives. tap is distributed as the open-source npm package @hua-labs/tap (v0.5.2).

End-to-end (E2E) autonomous-driving planners trained by imitation are prone to statistical shortcuts: they associate scene elements that merely co-occur with expert actions (a roadside object, a building facade) with driving decisions, rather than the variables that causally determine them. Such causal confusion silently compromises reliability in long-tail scenarios, and it is difficult to detect, because prevailing open-loop metrics (L2 displacement and collision rate) are dominated by ego status and do not indicate whether a planner depends on spurious cues. Existing remedies based on causal-intervention training require retraining large models and cannot audit a planner that is already deployed. We present CADET, a training-free framework that audits, benchmarks, and repairs spurious reliance in pretrained E2E planners without any parameter update.

How do people cope when disaster strikes and can we detect it at scale, in real time, from what they write? This study addresses that question using over one million Turkish-language tweets posted in the aftermath of the February 6, 2023 earthquake in Turkiye, which unfolded in a deeply polarized political context just months before a national election. Drawing on Lazarus and Folkman's (1984) coping theory, we develop a multi-label BERTurk classifier to detect three coping styles (problem-focused, emotion-focused, and meaning-making) across four theoretically motivated crisis phases. BERTurk achieves a macro F1 of 0.693, substantially outperforming a zero-shot mDeBERTa baseline (macro F1 = 0.324). Applied to the full corpus, the classifier reveals a clear temporal trajectory: problem-focused coping dominates the urgency phase and declines sharply, emotion-focused coping rises and stabilizes, and meaning-making increases monotonically. Anger correlates most strongly with meaning-making (Spearman r = 0.387), suggesting it functions as a mobilizing force toward blame attribution rather than practical action. These findings demonstrate that coping theory can be reliably operationalized in real-world digital crisis data and that doing so can help humanitarian organizations tailor their responses to where a population actually is.

We introduce COMET (Causal Object-centric Model for Efficient Tree search), a model-based reinforcement learning algorithm that performs Monte Carlo Tree Search in a slot-structured latent space. COMET pairs a frozen unsupervised object-centric encoder with a transformer-based world model, in which actions are bound to objects through a novel action-slot fusion mechanism that is used in slot transition prediction. Policy and value heads use object-causal attention, modulating token interactions by learned per-slot relevance scores so that decision-making concentrates on task-relevant entities. COMET adds an explicit object-level inductive bias to MuZero-style latent planning. Across eight visually and dynamically diverse tasks from the Object-Centric Visual RL benchmark, ManiSkill, Robosuite, and VizDoom, COMET achieves a higher mean normalized score during the early stages of training compared to object-centric and monolithic baselines.

Federated learning (FL) often struggles with generalization due to heterogeneous client data. Local models are prone to overfitting their local data distributions, and even transferable features can be distorted during aggregation. To address these challenges, we propose FedCONST, an approach that adaptively modulates update magnitudes based on the parameter strength of the global model. This prevents over-emphasizing well-learned parameters while reinforcing underdeveloped ones. Specifically, FedCONST employs linear convex constraints to ensure training stability and preserve locally learned generalization capabilities during aggregation. A Gradient Signal to Noise Ratio (GSNR) analysis further validates the effectiveness of FedCONST in enhancing feature transferability and robustness. As a result, FedCONST effectively aligns local and global objectives, mitigating overfitting and promoting stronger generalization across diverse FL environments, achieving state-of-the-art performance.

Safe reinforcement learning (Safe RL) aims to maximize expected return while satisfying safety constraints, typically modeled as Constrained Markov Decision Processes (CMDPs). While primal-dual methods scale well to deep RL, they often suffer from delayed constraint correction, leading to oscillatory behavior and prolonged safety violations. In this paper, we propose Constraint-Sensitive Policy Optimization (CSPO), a first-order primal-dual method that incorporates local constraint sensitivity into policy updates. CSPO augments the primal objective with a constraint-sensitive correction derived from the shortest signed distance to the safety boundary, enabling smarter recovery steps back to safety, compensating for delayed Lagrange multiplier updates, reducing oscillations near the boundary, and preserving the KKT solutions of the original constrained problem. Experiments on navigation and locomotion benchmarks demonstrate that CSPO achieves faster safety recovery and high reward preservation, resulting in higher constrained returns compared to state-of-the-art primal-dual and penalty-based methods

In this report, we present Hy-Embodied-0.5-VLA, abbreviated as HyVLA-0.5, an end-to-end system that spans the full robot learning stack: data collection, model design, continued pre-training and supervised fine-tuning, RL post-training, and real-world deployment. Each component serves a distinct role in this stack.

Mixture-of-experts (MoE) layers enable the scaling of transformer models while keeping the inference compute fixed. While task-expert specialization has been observed in empirical studies of frontier MoE transformer models, existing theoretical work analyzes this using continuous mixture models that cannot be used to model natural language effectively. An important open question is to \textit{theoretically explain task-expert specialization in transformer MoE models using discrete models of language}. To address this, we represent structured knowledge via syntactic templates and finite key-value dictionaries, and prove formally that a single-layer MoE transformer can encode knowledge by using experts that specialize in the corresponding tasks. Our construction shows how queries are routed to unique, task-specific experts whose size depends solely on the intrinsic complexity of the given task (i.e. the combined size of its syntactic templates and factual dictionary). Our construction provides a theoretical support for empirical results on localized knowledge circuits in MoE models. We support our theoretical findings with experiments evaluating model performance under varying MoE loss functions.

As agentic systems continue to evolve and are widely deployed in real-world scenarios, there is a growing demand to faithfully evaluate their capabilities. However, current benchmarks are typically built on popular applications with relatively simple tasks and focus on a narrow set of capabilities while overlooking broader dimensions, resulting in saturated performance on modern agents and failing to probe their limitations. To this end, we introduce GauntletBench, a web-based benchmark for evaluating agent generalisation in challenging scenarios, focusing on three underexplored capabilities (temporal perception, graphical understanding, and 3D reasoning), across five less-covered professional applications (Video Editor, Workflow Builder, 3D Modeller, Flight Analyser, and Circuit Designer), each with 20 vision-intensive tasks (100 in total). Our benchmark provides a modular pipeline that comprises an environment compatible with both open- and closed-source agent frameworks, a controlled web-based application, a well-structured task suite, and an automated evaluation engine with diverse metrics. Contrary to widespread expectations, our empirical results reveal that frontier agentic systems remain far from achieving human-level performance. Even the state-of-the-art agent achieves only a 19.1% success rate on our GauntletBench, highlighting the limitations in these overlooked capabilities and generalisation. By comparison, non-expert human annotators achieve over 80% success on our challenging yet feasible tasks, revealing the substantial gap between current agent capabilities and those required for complex real-world scenarios.

Despite advances in large-scale Automatic Speech Recognition (ASR), disfluent speech remains challenging, as state-of-the-art systems are often optimized to omit disfluencies, leading to information loss and hallucinations. Prior work has focused on verbatim transcription and the integration of disfluency markers, but adapting models on limited datasets can lead to catastrophic forgetting of general-domain knowledge. We address this gap by leveraging continual learning (CL) with explicit disfluency tokens. We first introduce these tokens into a pretrained ASR model to establish stable token mechanisms, and then continue training on additional datasets with varying disfluency distributions. Through a detailed analysis of model dynamics during training, we identify a trade-off between marker learning and ASR performance, and a consistent cross-attention head mechanism shared across CL methods.

Interventions designed to modify a particular behavior in LLMs, such as refusal or sycophancy, often produce unintended changes in other behaviors. This lack of targeted control makes it difficult to design and implement reliable safety controls. To understand these side-effects, we introduce a diagnostic framework for analyzing interacting behaviors in LLMs. We model behaviors as low-rank subspaces in activation space, and study how interventions influence across behaviors. Across multiple instruction-tuned models (7B-70B) and across refusal, jailbreak, and sycophancy settings, we find that different behaviors share internal representations, and intervening on one behavior alters others in asymmetric ways. Some behaviors act as upstream control points whose interventions propagate broadly across other behaviors, while others remain more isolated. We relate these effects to two geometric quantities: (i) the overlap between behavior subspaces, measured as the average squared cosine of principal angles, and (ii) the angle between each behavior subspace and the decision subspace (capturing the model's final decision e.g., refuse vs. comply). Empirically, intervention effects on other behaviors tend to be larger for behavior pairs with higher subspace overlap, and for source behaviors whose subspaces lie closer (smaller angle) to the decision subspace. These findings highlight a challenge for targeted behavior control: behaviors are difficult to modify independently, as interventions can propagate through shared representations and asymmetric interactions.

Scientific discovery saturates when new hypotheses cease to provide independent information, even if the nominal hypothesis space remains large. We study hybrid discovery systems that combine structured local search with LLM-generated non-local proposals and pose the Search Compression Hypothesis: non-local exploration helps only when three geometric conditions co-occur: spectral compression, orthogonal escape from the explored span, and residual signal alignment with the target. We formalize these conditions, derive necessary conditions for hybrid advantage, and test the mechanism in controlled synthetic environments, large-scale A-share factor discovery, and symbolic-regression benchmarks; a public tabular operational sanity check tests the associated budget-allocation implication. Signal-planting and directed-versus-random experiments show that novelty alone is insufficient: random orthogonal jumps expand coverage but do not improve yield without predictive alignment. Across compression sweeps, real factor archives, and LLM-SRBench tasks, hybrid gains concentrate in weakly represented but target-bearing directions and vanish as the hypothesis space approaches full rank. The framework turns LLM-guided discovery from generic novelty search into a diagnostic procedure for deciding when directed non-local exploration is warranted.

Vision-language-action (VLA) models are powerful action generators for robot manipulation, but they are typically executed with fixed inference and replanning schedules. This rigidity ignores the uneven difficulty of robot control: contact-rich or uncertain states may need more computation and fresher feedback, while easier states can often be handled with fewer inference steps and longer open-loop execution. We propose Elastic Queries Reinforcement Learning (EQRL), a framework that makes each VLA policy query elastic. A lightweight latent-schedule adaptor jointly selects the latent input, denoising budget, and action chunk length, without fine-tuning the underlying VLA model. To make scheduling difficulty-aware, EQRL trains a critic over the joint latent-schedule action and derives a state difficulty signal from critic ensemble disagreement. This signal guides compute toward difficult states, while a learned residual allows task-driven correction. We formulate variable chunk execution as query-level macro-action RL with chunk-dependent discounting and an amortized number-of-function-evaluations (NFE) budget. Across simulation and real-robot manipulation, EQRL reduces amortized inference cost while preserving or improving task success.

The workflow from particle collision to physics analysis passes through a series of reconstruction steps that are traditionally modular and disconnected, with no shared representation linking low-level detector data to high-level analysis tasks. We show that casting event reconstruction as a machine learning problem naturally produces such a shared representation. We repurpose a machine learning model trained for particle-flow reconstruction (MLPF) to perform three distinct analysis tasks: jet flavor identification, jet energy regression, and missing momentum regression. By appending the per-particle latent representations learned during reconstruction as additional input features, we substantially improve over baselines that use kinematic features alone. We further demonstrate that a single linear layer trained using only the latent representations achieves competitive performance against state-of-the-art baseline architectures, and outperforms the baseline for missing momentum regression with approximately 35 times fewer parameters. These results demonstrate that the latent representations learned during reconstruction encode essential physics information needed for downstream analysis, establishing MLPF as a foundation model and offering a concrete step toward an end-to-end pipeline from detector data to physics analysis.

We study multi-domain LLM training in which two models, each stronger in a different domain, co-evolve by tutoring each other through on-policy feedback. Unlike one-way distillation or single-model fine-tuning, our goal is mutual Pareto improvement: each model improves across domains without losing its original strength. To this end, we propose On-Policy Co-Distillation (OPCoD), where each student's self-distillation is conditioned on its own correct rollout and feedback from its peer. To make feedback exchange effective, OPCoD uses cognizance-based gating to decide when to give feedback and feedback anchoring to ground feedback in the problem. On Science Q\&A tasks, OPCoD consistently outperforms baselines and achieves Pareto improvement across all evaluated domain pairs and students.

Machine learning (ML) pipelines require extensive data preparation, feature engineering, and integration across heterogeneous sources, making them tedious and error-prone to develop. While large language models (LLMs) have recently shown promise for assisting programming tasks, chat-based interfaces provide limited control over pipeline behavior and often produce code that is difficult to optimize or integrate into production systems. We demonstrate SemPipes, a novel programming model that extends ML pipelines with declarative, LLM-powered semantic data operators. SemPipes allows developers to specify high-level natural language instructions for data-centric operations, while seamlessly combining these operators with arbitrary Python code from standard data science libraries. For the semantic operators, it synthesizes specialized implementations at pipeline training time, conditioned on dataset characteristics and pipeline context, enabling the flexible yet controlled integration of LLM capabilities. We demonstrate SemPipes through SemPiper, an interactive interface that visualizes computational graphs of the pipelines, synthesized operator implementations, and optimization trajectories produced by an evolutionary search procedure. Attendees can explore three end-to-end scenarios, modify pipelines, inspect generated code, and observe how semantic operators are synthesized and iteratively optimized. The demonstration highlights how declarative semantic operators enable controllable, optimizable, and practical integration of LLMs into ML pipeline development.

Frontier coding models may spend substantial capacity learning not only program behavior, but also accidental entropy in human repositories. Such repositories contain valuable signals: tests, incidents, migrations, edge cases, product judgment, and operational history. These signals are entangled with framework churn, naming drift, generated-source ambiguity, dependency rituals, CI dialects, weak proof routes, and human-oriented review customs. We propose agent-first canonical code, a proof-carrying substrate that rewrites routine product software into canonical behavior profiles, typed change algebra, proof lanes, constrained edit grammars, semantic patch cells, runtime negative memory, and proof-carrying change objects. The core hypothesis is that quotienting software by behavior equivalence under a declared oracle can collapse equivalent encodings into governed representatives with explicit evidence and proof obligations. The endpoint is amortized cost per verified correct change, including source, context, reasoning, tools, verification, security, provenance, review, failed loops, defects, and foundry cost under a common oracle. Reported reduction bands are hypotheses, not measured frontier results. The proposed limit is a No-Accident Horizon: removable accident decreases until residual novelty, evidence, governance, risk, and future optionality dominate. For supported routine-product distributions, this gives a defensible planning target near 100-fold all-in cost reduction, not a guarantee for all software. Preliminary QLoRA experiments on Qwen2.5-Coder-14B show that 64,088 canonical trajectories are learnable and suppress tested forbidden-language markers, but do not establish behavior preservation, scaling economics, or verified-change cost. The contribution is a falsifiable program centered on minimum functional description length and verified-change cost.

Applications in the 3D Computing Continuum, which unifies edge, cloud, and space, require combining multiple AI tasks such as object detection, time-series analytics, and natural language processing into Compound AI systems. These systems must satisfy stringent Service Level Objectives (SLOs) on accuracy, latency, and cost. A key mechanism for maintaining SLO compliance of Compound AI systems is runtime model selection, where AI models are dynamically switched for each workflow task. However, existing distributed and compound AI frameworks do not natively support runtime model selection. We present PLAIground, a framework that enables runtime model selection for Compound AI systems. PLAIground introduces Compoundable AI Model (CAIM) abstraction, which decouples task semantics from AI model implementations via Task and Data Contracts, enabling model switching without workflow changes. Additionally, PLAIground introduces Pixie, an SLO-driven runtime model selection algorithm, which dynamically selects the most suitable model for each task during execution. Our evaluation on two realistic Compound AI workflows demonstrates that Pixie achieves up to 91.3% accuracy while maintaining SLO compliance where fixed-model strategies either violate cost and latency budgets up to 21x or miss accuracy targets by 4%.

Federated learning is well suited to edge environments but is often limited by the uplink cost of transmitting model updates. This Work-in-Progress paper presents MUFFLe, a communication-efficient update compression scheme that integrates generalized deduplication (GD) into the FedAvg pipeline. MUFFLe deduplicates repeated patterns across the update vector, yielding a fixed-rate, variable-count compression scheme. Preliminary experiments on IID MNIST with 20 clients show that MUFFLe reaches the target accuracy of $92.93\%$ with 38~MB cumulative uplink communication, compared with 75~MB for 8-bit quantization, 86~MB for Top-$k$ sparsification, and 310~MB for uncompressed FedAvg. These results demonstrate the feasibility of applying GD to communication-efficient federated learning.

Error-bounded lossy compression is a fundamental technique for managing the rapidly growing volumes of scientific data produced by modern simulations and observational instruments. Most state-of-the-art-compressors follow a prediction-residual paradigm, where compression effectiveness depends on the quality of the predictor: more accurate predictions generate smaller residuals that are easier to compress. This observation raises a question: can modern machine learning models serve as superior predictors for scientific data compression? Answering this question directly is challenging because developing compression-specific ML predictors requires substantial resources. Instead, we leverage the climate domain where highly accurate pretrained weather forecasting foundation models already exist, making them an ideal testbed. We present a framework that integrates spatial and temporal deep learning models into a conventional error-bounded compression pipeline. The framework supports auto-regressive forecasting models and avoids error accumulation. Using ERA5 climate data as a representative large-scale scientific dataset, we evaluate three distinct ML predictors: a VAEformer-based codec (CRA5), a graph neural network forecaster (GraphCast), and a vision-transformer forecaster (Aurora), against the state-of-the-art compressor SZ3.1 under identical quantization and entropy-coding backends. Our evaluation over approximately 1.7 TB of data reveals a surprising result: although ML predictors generate more accurate predictions and can improve reconstruction quality by up to 91% while achieving up to 9.6x higher compression ratios for highly predictable variables, they do not improve overall dataset-level compression ratio. We show that prediction accuracy alone is insufficient: the spatial structure of the resulting residuals plays a decisive role in entropy coding efficiency.

Artificial Intelligence (AI) systems must typically satisfy service-level objectives including accuracy, latency, and cost. The prevailing model-centric approaches select a monolithic model at design time and apply identical computation regardless of input difficulty, cannot decompose tasks across specialized components, and have knowledge that is fixed at training time. During runtime, this can lead to performance degradation and increasing costs. Because the model is the main design variable, it determines the majority of system behavior, coupling operational objectives to a single design-time choice. Addressing these limitations requires shifting from model-centric to system-centric design. Compound AI systems realize this shift by orchestrating multiple models, algorithms, and tools as distributed AI systems through explicit control logic. The performance of such systems depends on their workflow topology, the models assigned to each task, and the parameters governing runtime behavior. We present a design methodology that organizes this space along two dimensions, workflow topology and configuration selection, and identifies eight design patterns, each consolidating techniques to address a specific limitation of monolithic deployment. We validate our methodology through three case studies. Across our case studies, Compound AI configurations approach accuracy of monolithic models within 2.5 to 4 percentage points while reducing latency by up to 60% and cost by up to 71%. We show that model selection and parameter configuration jointly determine system performance, but the resulting design space grows combinatorially, as workflows compose more patterns and components. Thus, we identify five open challenges that define a roadmap from manually configured prototypes towards systems that automatically discover and maintain SLO-compliance in Compound and Distributed AI systems.

The increasing availability of large-scale textual corpora has opened new possibilities for data-driven, quantitative approaches to historical analysis using Natural Language Processing (NLP). However, diachronic corpora with historical relevance from the pre-digital era remain scarce and often incomplete. We present a quantitative approach to historical analysis based on the reconstruction and exploration of a diachronic corpus of around 600,000 articles from the Italian newspaper "La Repubblica", covering all the articles published from the 1st of January 1985 to the 31st of December 2000 - a period of major political, social, and geopolitical change in Italy and globally. Using NLP techniques, we analyze the text at both lexical and semantic levels; we then apply tools from complex systems and statistical physics to trace shifts in media discourse over time. This allows us to detect key transition periods, such as the transition from the First Republic to the Second Republic in Italy, or major international conflicts like the Gulf War or the Kosovo War, without relying on prior labeling. The results show how combining computational linguistics with ideas from complex systems can offer new quantitative insight into historical changes, opening up new paths for studying the dynamics of media and society through large-scale textual data.

Large language models exhibit strong multilingual capabilities, however, their internal representations are difficult to interpret. Understanding these interactions is important for ensuring reliable behavior in multilingual systems. Recent work has shown that causal-geometric structure can explain how certain concepts are encoded as approximately linear and separable directions, but whether this framework extends to multilingual models, where language identity is correlated and hierarchical, is underexplored. We apply causal-geometric analysis to multilingual LLMs, studying 28 bilingual contrasts across three models, allowing us to analyze when languages behave as approximately independent factors and when structured dependencies persist. We find evidence that language concepts admit stable linear representations that are largely separable under a covariance-adjusted (causal) inner product, with structured deviations reflecting linguistic similarity. Moreover, languages within the same family (such as Germanic or Romance) exhibit a simplex-like geometric structure, suggesting hierarchical organization. These results extend causal-geometric interpretability to multilingual settings and provide insight into how separability and similarity may exist in multilingual LLM representations, motivating interpretability analyses that diagnose when and how structured dependencies between concepts can be anticipated. This has implications for trustworthy deployment, as residual structure between languages may lead to unintended cross-lingual effects when models are monitored or intervened upon.

Unstructured pruning produces sparse weight tensors, but the standard implementation keeps tensor shapes unchanged so the deployed model is no smaller than before pruning. We present an exact structural rewrite, which we call minimization, that converts a masked network into a smaller dense network with the same forward function up to floating-point rounding. The Squeeze-Release cycle iterates pruning and minimization with an intermediate release step that re-enables the exact-zero positions inside the compacted tensors as small calibrated noise, turning otherwise wasted capacity back into trainable parameters. Successive cycles use that capacity to find structural redundancy a single pass cannot reach. We additionally introduce CompensatedLayerNorm, a function-preserving replacement for LayerNorm that extends minimization to channel reduction across LayerNorm-equipped residual streams. Squeeze-Release compresses the deployable network to 39x smaller than the unpruned model on a fully-connected model network and 14.8x smaller on modern CNN (ConvNeXt-Tiny), at comparable accuracy. In addition we prove that the rewrite can be extended to transformer architectures.

Tactile imaging seeks to reconstruct the internal structure of soft objects through touch sensing, with applications in medical diagnosis and robotic manipulation. Recent self-supervised learning approaches have shown promising results, but rely on global, unstructured representations and robot-controlled sensing, limiting generalization and practical use. We propose Local Encoder for Spatial Sensing (LESS), an object-centric tactile representation that exploits the local nature of touch. The tactile scene is modeled as a grid of recurrent encoders with local receptive fields, whose states are fused to reconstruct 2D or 3D images of internal structure. This compositional design enables strong generalization: models trained on single-inclusion phantoms accurately image objects with multiple inclusions and varying sizes. The local structure further supports spatial uncertainty estimation. In addition, we enable hand-held tactile imaging via external pose tracking and human-like palpation data, and extend tactile imaging to full 3D reconstruction.

Recovering structural information from noisy high-dimensional data is a fundamental task in statistical inference. We investigate the recovery thresholds for a graph hidden in a randomly weighted complete graph. Specifically, an unknown graph $H^* \in H_n$ is chosen uniformly at random, and hidden in a complete graph of $n$ vertices as follows: the weight of an edge $e \in H$ is distributed independently according to $P_n$; otherwise the weight is distributed independently according to $Q_n$. The goal is to recover almost all of $H$ from these edge weights. Assuming a local Lipschitzness of the Rényi divergence between distributions $P_n$ and $Q_n$, and a mild density condition for the graphs $H_n$, we give a unified characterization of the information-theoretic limit for recovering almost all of $H$ (also known as almost exact recovery). Our characterization connects the KL divergence between $P_n$ and $Q_n$ to the logarithm of the first moment threshold of $H$ in the Erdős-Rényi random graph model $G(n,p)$. Our lower bound also extends to the task of partial recovery, in which only a constant $λ$-fraction of $H$ needs to be recovered. Last but not least, for certain Bernoulli and Exponential regimes, and for Gaussian distributions, we are able to show an All-or-Nothing (AoN) threshold phenomenon at the exponential scale.

High-dimensional datasets often concentrate near low-dimensional structures, but estimating their geometry from samples typically relies on graphs and kernels that scale poorly with dataset size and dimension. We propose Riemannian metric matching: a denoising probabilistic framework for learning the Riemannian geometry of data using neural networks. Specifically, we learn the carré du champ operator, which, using diffusion geometry, gives us access to the Riemannian geometry toolkit for downstream machine learning and statistical tasks. Our key observation is that the carré du champ operator can be formulated as a conditional expectation over random perturbations of the data, which can be exploited for sample-wise training and constant cost, amortized inference without explicit kernel construction. Empirically, metric matching rivals or improves the accuracy of $k$-NN-based diffusion geometry estimators, while enabling amortized inference that is up to $400\times$ faster, and supports graph-free geometric analysis on high-dimensional images where nearest neighbors break down.

This paper appraises recent frameworks within AI development to integrate LLMs into control tasks in automotive contexts from the perspective of safety assurance. This work has built upon the rapid integration of LLMs across automotive settings. However, we find that at present, these frameworks face significant challenges, limiting their efficacy in real-time safety-critical contexts. Firstly, we consider conceptual challenges, including the fact that deployers are faced with a dual challenge, wherein they must assure a model which has been developed upstream, i.e. as general-purpose tools by the large AI labs, in a downstream context, i.e. into specific vehicle architectures. Secondly, we consider concrete challenges from across existing standards. We show that there are currently both fundamental engineering constraints covered in ISO21448, such as latency, and novel LLM-specific issues, such as alignment-related issues covered in ISO/PAS8800. We ground both examples in a concrete introductory, experimental case study exploring an existing open-source repository, Talk2Drive. We present a safety argument in order to make explicit the limitations of existing solutions. Nonetheless, given that the use of LLMs in automotive contexts is being explored at a technical level and operationalised, we propose potential assurance mechanisms for LLM-related hazardous events going forward.

Property Graphs are rapidly being adopted as database frameworks for representing heterogeneous data sources. To enable precise access to the information contained in them we need conversational interfaces based on Text-To-Cypher (Text2Cypher) parsers. This paper presents an automatic synthetic data generation method that can be leveraged to fine-tune small LLMs for this task. We conduct experiments on all the major Text-To-Cypher benchmarks, demonstrating that with our synthetic data generation approach we can significantly increase the performance of small LLMs, allowing them to compete with much larger proprietary models. This means that in settings in which models must be locally deployed we can ensure data-sovereignty without sacrificing accuracy and without costly annotation campaigns.

LLM agents have rapidly evolved into autonomous systems, yet a persistent information gap remains between users and agents: communication is costly, while users' identical preferences further limit information exchange. To investigate how agents should communicate across modalities, this paper formalizes Communication Policy, establishes textual and UI-based policies, and then evaluates communication policies across diverse environments, personas, and model combinations. Building information asymmetry for proactive agents, we set up two complementary settings, User-Agent and Planner-Executor. Experimental results reveal complementary strengths between interaction channels: text-based interaction often facilitates task performance, while structured UI improves agents' response quality and persona compliance. Motivated by that, a hybrid method combines these advantages. We further propose Communication Policy Evolution (CPE), a self-evolution framework for refining communication policies through rollout and prompt-level evolving. Without model modification, CPE achieves the best task success across multiple settings using prompt refinement alone. Our findings identify communication behavior as a critical yet underexplored design dimension for LLM agents.

Real-world spatio-temporal forecasting must handle irregular time points, spatially sparse observations, and the need for uncertainty quantification. This setting is often further compounded by nonlocal interactions (long-range spatial coupling). Modeling continuous-space, continuous-time nonlocal dynamics naturally leads to infinite-dimensional integro-differential equations (IDEs), making principled Bayesian inference intractable. We propose the NonLocal Bayesian Spatio-Temporal model (NLBST), a hierarchical Bayesian framework for continuous spatio-temporal fields that learns explicit nonlocal coupling while retaining tractable inference. NLBST represents the latent field via a coordinate-based spatial basis expansion and models the coefficient process with a continuous-time ODE whose learnable linear operator corresponds to a Galerkin reduction of a nonlocal IDE; a Neural ODE residual captures additional nonlinear dynamics. A linear-Gaussian observation model enables Kalman-style sequential updates under missing and irregular observations, while the spatial basis representation enables inductive prediction at unmeasured locations without retraining. Global parameters are learned via variational inference, and uncertainty is handled through a Bayesian hierarchy. Experiments on synthetic and real-world datasets demonstrate strong forecasting and spatial generalization with well-calibrated uncertainty, yielding substantial gains over baselines in strongly nonlocal and partially observed regimes.

Property graphs may be constrained by schemas that inform both query engines and human users about the shape of valid data, enforcing a contract between data provider and consumer. Composable property-graph queries transform input graphs into output graphs. Then, the question arises of which schema can be expected after one (or several) transformation steps. We investigate how schema constraints can be inferred given an input schema and a transforming query. Specifically, we propose a reasoning procedure that, given an input schema in ProGS and a query in G-CORE infers an output schema. Since graph updates will happen frequently, our inference procedure does not rely on graph instances, such that the computed output schema applies to all graphs originating from any input graph complying with the input schema. Related work has addressed this problem for SPARQL CONSTRUCT queries, encoding it in Description Logics (DLs) so that the output schema is entailed by axioms inferred from input schema and queries. Property graphs and their queries, however, complicate the matter, as property graphs feature label and property annotations as well as first-class edges. Thus, reification has to be used in one way or another, though available DLs lack the means to encode such features directly. We approach this novel challenge via a family of mappings for i) property graphs reified in RDF, aligned with ii) a mapping from ProGS to SHACL and iii) a mapping from G-CORE to SPARQL CONSTRUCT queries. In this manner, schema inference for property graphs becomes manageable, as we break apart the problem through the extra mapping layer and utilize efficient DL reasoners. We develop the metatheory regarding the soundness of inferred schema constraints and the semantic equivalence of mapped schemas and queries.

Current Generative AI (GenAI) interfaces remain largely constrained to chatbox interaction, which can impose high cognitive demands on users and create substantial barriers for people with intellectual disabilities (ID), including prompt formulation difficulties, response overload, and limited mechanisms to assess information reliability. To explore alternative interaction models for cognitive accessibility, we conducted a cross-disciplinary co-design challenge in which two student cohorts (Computer Science and Industrial Design) developed interface concepts from the same set of functional requirements (e.g., prompt scaffolding, structured output, GUI-based refinement, transparency, and personalization). Comparing the resulting proposals reveals both convergence on foundational requirements (notably initial calibration, proactive prompting, and direct manipulation of response fragments) and complementary contributions that outline a multi-layered support system. Computer Science teams primarily produced structural scaffolding, emphasizing predictability, navigability, and trust through mechanisms such as reliability indicators, explicit sources, and context management for long conversations. Industrial Design teams emphasized experiential scaffolding, focusing on pacing, attention guidance, multimodality, and proactive agency, including step-by-step response flows, focus modes, and assistant-like integrations. We synthesize these findings into a dual-layer scaffolding framework that expands the design space for cognitively accessible GenAI interaction beyond chat-centric models and motivates future work on expert refinement, technical feasibility, and empirical validation with users with ID.

LLM-based agents trained with reinforcement learning optimize step-wise action prediction but lack metacognitive awareness of task progress, inducing a gap that hinders long-horizon scaling. A pilot study reveals that online progress prompting hurts performance while retrospective demonstrations help, yet this capability cannot emerge from outcome-reward training alone. We present RePro, Retrospective Progress-Aware Training, a framework that trains agents to self-generate progress signals via a forward-then-reflect rollout paradigm: the agent executes actions online, then retrospectively reassesses its step-wise progress given the completed trajectory and known outcome. RePro initializes with a Retrospection Warmup that teaches reflection format from minimal external demonstrations, then further trains through RePro-PO with a composite reward that produces self-generated signals without continuous external supervision. Experiments on WebShop, ALFWorld, and Sokoban show that RePro enhances the Qwen family's performance, with up to $12\%$ absolute success rate gains.

Vision-Language Models (VLMs) such as CLIP have become a standard backbone for open-vocabulary recognition, yet their zero-shot predictions remain vulnerable to distribution shifts encountered at deployment. Test-Time Adaptation (TTA) has recently been extended to CLIP as a lightweight solution, leading to a rapidly growing body of TTA4CLIP methods. However, empirical progress in this area has largely outpaced our understanding of what truly drives adaptation, where their gains originate, and under which shifts they remain reliable. In this paper, we take a step back from the pursuit of state-of-the-art accuracy and conduct a systematic controlled study of TTA4CLIP. We first organize existing methods into three unified paradigms according to what is updated at test time. We then introduce TTABC, an open-source TTA Benchmark for CLIP, which standardizes evaluation protocols and integrates more than 20 representative methods. Our controlled empirical analysis focuses on three key areas. First, we determine the driving factors in parameter-based methods, revealing that adaptation gains are primarily driven by test-time evidence and reliable proxies rather than heavy optimization. Second, we explore evidence utilization beyond heavy parameter tuning, showing that competitive and efficient performance can be achieved through cross- or current-sample evidence and lightweight prototype updates. Finally, we demonstrate that there is no silver bullet for TTA: no single adaptation paradigm is universally optimal, and the preferred paradigm depends on the nature of shift. We hope our benchmark and study provide a clearer understanding of the current TTA4CLIP landscape and establish a foundation for further research.

Developing accurate crowd-counting models for Hajj pilgrimage scenes remains challenging because domain-specific annotated images are scarce and data collection during large gatherings raises privacy concerns. To address these limitations, this paper proposes Pix2Pix-Hybrid (P2P-H), a hybrid conditional GAN for structure-guided Hajj crowd-image synthesis and data augmentation. P2P-H builds on Pix2Pix and employs a U-Net generator conditioned on eight input channels that jointly encode structural cues (edges and grayscale) and contextual attributes (crowd density and time of day). To capture detailed textures in dense scenes, the framework integrates two multi-scale PatchGAN discriminators operating at different resolutions. The training procedure combines adversarial, perceptual, and feature-matching objectives with adaptive data augmentation and stabilization strategies. The model was trained on 993 real Hajj frames collected from 60 publicly available video sources, with conditioning attributes derived automatically to reduce manual labeling effort. Using this framework, we constructed CrowdH, a synthetic dataset of 10,000 high-resolution Hajj crowd images. Experimental results show that P2P-H improves structure-preserving conditional synthesis quality compared with Pix2Pix and StyleGAN2-ADA baselines and shows favorable transfer to other crowd datasets. To assess downstream utility, we further constructed CrowdH-Mix-469, an annotated mixed real-synthetic dataset comprising 384 real Hajj images and 85 selected synthetic images,and evaluated five crowd-counting models under real-only and real-plus-synthetic training. The selected synthetic data reduced MAE across all five models, with the strongest gain observed for CSRNet.

Frontier AI systems are increasingly capable of cybersecurity tasks, including codebase inspection, vulnerability detection, and exploitation. However, evaluating their offensive capabilities remains constrained by limited access to open, reproducible, multi-host cyber ranges. Existing public benchmarks capture isolated skills such as CTF solving, vulnerability reproduction, and exploit generation, but often abstract away realistic intrusion workflows: discovering exposed services, gaining a foothold, collecting internal information, and expanding compromise across hosts. This gap makes it difficult to observe emerging risks early, because frontier AI systems are rarely evaluated under realistic attack conditions. We introduce AgentCyberRange, the first open, multi-range infrastructure for measuring autonomous cyber attack capability in realistic cyber ranges. It combines 110 vulnerabilities across 15 real web applications and 8 enterprise-like cyber ranges with 156 internal hosts, plus Cage, a toolchain for execution, orchestration, result collection, and verification. The benchmark covers two core stages: web exploitation, where agents explore exposed applications and validate vulnerabilities, and post exploitation, where agents turn an initial foothold into broader internal compromise. We evaluate six frontier AI systems under matched prompts and budgets. GPT-5.5 with Codex performs best, solving 16.1% of web exploitation tasks and 31.7% of post-exploitation tasks; with more concrete hints, these rates increase to 33.0% and 46.3%. We also observe out-of-benchmark findings, including unknown vulnerabilities in popular projects, and payload mutation that bypasses host defenses. These results show that open cyber-range evaluation is necessary for observing emerging offensive capabilities under realistic and reproducible conditions.

Population-based and distributional optimization methods, from evolution strategies and consensus-based optimization to covariance-matrix adaptation and stochastic gradient methods viewed as distributional dynamics, are widely used for nonconvex or black-box problems, yet their convergence analyses remain fragmented across algorithm-specific techniques. We introduce an operator calculus in which a broad class of such methods, after choosing an appropriate state space and, where necessary, augmenting the state by memory or strategy variables, is described as a composition of three elementary operators (mutation, selection, and recombination) acting on probability measures. Under explicit stability and regularity conditions, the composite operator admits a pre-generator whose continuous-time limit is a transport-reaction-jump (TRJ) PDE that preserves the operator splitting. On this foundation we establish a modular Lyapunov principle. If a state-space Lyapunov function both dissipates under the full generator and controls the relevant search-space gauges, then the state-space Lyapunov functional and the induced search errors decay exponentially. The additive generator structure allows dissipation estimates to be assembled operator by operator, providing a toolkit for certifying convergence of composite mean-field algorithms.

Time series prototype learning is fundamentally challenged by observational ambiguity. Discrete architectures fail to resolve this, as they lack the capacity to decouple stochastic noise from continuous dynamics. Furthermore, rigid closed-set assumptions fail to capture unseen diversity. To address these limitations, we propose a hierarchical ordinary differential equation clustering network, which utilizes neural ordinary differential equation to model latent state evolution as a continuous integral curve. This formulation enforces temporal continuity to effectively disentangle smooth feature trends from stochastic noise, while our adaptive hierarchical mechanism autonomously determines the appropriate number of prototypes without rigid prior constraints. Validated on the early link failure detection task with irregularly sampled time series, the proposed method effectively extracts underlying physical prototypes, thereby enabling robust failure detection. Our code is available at https://github.com/NJ-LNN/Hierarchical-ODE.

Deep learning has driven many recent advances in process analytics, especially for predictive and prescriptive monitoring. However, standard objectives such as cross-entropy optimize local next-step likelihoods and only implicitly capture control-flow structure. As a result, models can achieve high token-level accuracy while permitting imprecise global behaviour. We introduce DIFF-ERO, a conformance-aware loss function for deep learning models on process data. DIFF-ERO is a differentiable formulation of entropy-based stochastic conformance that incorporates control-flow information during training. Our approach constructs batch-level stochastic transition matrices with soft edge memberships, allowing structural precision and recall signals to directly inform backpropagation. The loss is model-agnostic and can be applied whenever the final representation parametrizes stochastic transitions. We instantiate DIFF-ERO in transformer encoder-decoder pipelines for next-activity prediction and use it jointly with cross-entropy to analyse its theoretical components with respect to convergence. Across benchmarks comparing other loss functions and targets, DIFF-ERO shows improved predictive performance where structure matters most while maintaining parity elsewhere. At the same time, the learned stochastic automaton converges towards the structural ground truth, indicating that the network internalizes process model structure.