The Inference Report

Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterogeneous supervision according to the dependencies among generative capabilities. We present a \textbf{capability-driven data infrastructure} that couples capability-specific supervision construction with capability-aligned curriculum scheduling. Its three specialized yet interoperable data engines build complementary relational supervision for text-image grounding, inter-image transformation, and image-knowledge association, while caption experts align T2I and editing supervision across tasks and granularities. A multi-stage curriculum jointly evolves task composition, visual-concept distribution, data quality, and image resolution along the dependency order of capability acquisition, with capability-aware evaluation closing the loop through targeted retrieval, expert construction, and gap-aware resampling. At scale, the framework curates a 440M-image T2I corpus, 120M editing pairs, and over 27M image-entity pairs. With this infrastructure, we train multimodal diffusion models at two scales from scratch, with 3B and 6B sizes respectively. We conduct quantitative evaluation on CPI-Bench, along with qualitative evaluations across diverse text-to-image and editing scenarios. Experimental results present broad visual coverage, versatile rendering, and effective transfer across generative capabilities.

Purpose: To develop and evaluate a locally deployed multi-agent AI system for radiology report structuring and quality assurance. Materials and Methods: This retrospective study included 638 radiology reports from CT examinations of the chest, abdomen, and pelvis dictated by 15 board-certified radiologists in 2023 and 2024. A multi-agent AI pipeline was developed to perform report structuring and quality assurance (QA). The system structured the report into standardized anatomical sections at the sentence level using regex rules and local large language models. It also detected mismatches between the Findings and Impression sections, or within sections; gender-anatomy conflicts; and undocumented communication of critical findings. Two board-certified radiologists independently evaluated a 45-report subset. Results: The multi-agent system structured the Findings sections of all reports (22,270 sentences) into a predefined anatomical format while retaining the original report content. The system flagged 90 (14.1%) reports, most commonly for section mismatches (80 reports, 12.5%). In the radiologist evaluation, both reviewers agreed that 31 (69%) were correctly restructured, 2 reports (4%) were incorrectly restructured, and disagreed on the remaining 12 reports (27%). Both reviewers agreed that no clinically important information was omitted and no fabricated content was introduced. Overall QA performance was rated as "excellent" or "good" in 84% of the evaluated reports, with the remaining reports rated as "fair". Conclusion: A locally deployed multi-agent AI system combined radiology report structuring and quality assurance within a single workflow. The system demonstrated favorable performance in radiologist evaluation. Such systems may support standardization of reporting and quality assurance in radiology practice.

Memory-based self-improving agents--those that learn from an online stream of tasks and improve over time by maintaining a textual memory bank--have shown great promise in recent literature. However, the reliability aspects of these methods have been critically overlooked. In this work, we conduct a comprehensive re-evaluation of two memory-based methods, broadening the scope of evaluation along two axes: (1) including multiple runs to quantify variance, and (2) randomly shuffling the tasks to investigate the effect of task order. Through these experiments, we make two observations that expose the fragility of current methods: First, agent evaluation is inherently noisy in complex environments and on multi-step tasks, and stacking a self-improving loop on top can further amplify this noise. Second, the agent's improvement is highly dependent on task order. Prior works often adopt default orderings that impose an implicit curriculum, acting as a hidden prerequisite for success. To better understand this fragility, we manually examine the agents' memory and hypothesize that task and environment underspecification contribute to this fragility. We validate this hypothesis by incorporating information that enables better specification, such as detailed rubrics and environment feedback, into the memory construction process. While this added information partially closes the performance degradation in previous experiments, significant gaps still remain, suggesting that other uncharacterized factors contribute to this fragility. Looking ahead, our work advocates for more rigorous evaluation protocols for self-improving agents by reporting results across multiple runs and stress-testing them under challenging conditions. Moreover, our findings on underspecification call for systems and interfaces that enable effective human oversight, preventing agents from failing in unforeseeable ways.

Language model tokenizers are typically selected with minimal evaluation, despite the fact that their design choices directly impact model capabilities. This can be partly attributed to a limited understanding of which tokenizer properties affect which aspects of downstream performance. We introduce TokEval, a framework of tokenizer evaluation metrics that goes beyond standard measures like fertility and compression rate to capture linguistically and structurally meaningful properties, e.g., UTF-8 character boundary integrity and digit place-value boundary alignment for mathematics. To validate whether these metrics are predictive of downstream model performance, we conduct controlled language model pretraining experiments, varying solely the tokenizers' training data mixture, pretokenization strategy, and training algorithm. We evaluate the resulting models on bits-per-byte (a tokenizer-agnostic version of perplexity) and several benchmarks, spanning linguistic understanding, mathematical reasoning, and code generation. Our experiments suggest that different intrinsic properties have different impacts on model abilities: information-theoretic metrics predict language modeling abilities (Spearman rho up to 0.80), while structure-sensitive metrics, such as those measuring digit and line-break handling, correlate with task accuracy. We hope TokEval enables more principled tokenizer evaluation, replacing pretraining sweeps with intrinsic measurement wherever the two agree.

We give a two-player zero-sum repeated game between a learner and nature whose value identity generates Bayesian updating and an exact accounting of exponential-weights regret at once, and supplies the comparator-class variational form that a wide class of concentration phenomena share. The terminal payoff is the most a comparator can gain at fixed relative entropy from the prior, and the one-step constraint is an information budget on nature's move under the learner's mixed action. With the learner's move otherwise unrestricted, Gibbs/Bayes weights emerge as its unique Bellman equalizer -- the mixed action that makes the per-round loss independent of which direction nature moves -- with log-partition functions playing the role of value functions. The regret decomposes exactly into three parts: a per-round information loss reflecting the variation in observed outcomes, an additive retempering drift that accounts exactly for any change of measurement scale between rounds, and the information the comparator carries relative to the prior. The variance and bounded-range proxies that drive standard regret bounds are looser relaxations of this decomposition, which holds generally and governs them all. Both players' strategies are read off from the decomposition term by term, and repeated play yields an information-theoretic ledger of self-play in place of the usual quadratic-variation surrogate. The same comparator-class geometry accounts for the classical large-deviation bounds, and methods across bandits, posterior sampling, aggregation, and boosting are specializations of the one regret decomposition.

Autonomous LLM agents that converse on a user's behalf are an emerging design pattern in matching platforms, yet their viability depends on a condition rarely examined: users must accept not only delegating conversation to an agent, but also receiving agent-mediated communication from others. We study this condition using two large-scale surveys of active users of a major dating platform (N=2,894 on generative profile features; N=2,617 on autonomous conversational agents, fielded in two languages). We develop a latent-variable measurement model of agent receptivity based on graded response models with latent regression, and show via model comparison that willingness to send and willingness to receive agent communication are distinct constructs: highly correlated (rho=0.92) but separable (Delta BIC=52), with partial measurement invariance across languages. The model quantifies a systematic delegation asymmetry: deploying one's own agent requires far lower receptivity (threshold -0.38) than engaging a counterpart's agent (+0.32; full engagement +1.39), and mean deployment propensity exceeds engagement propensity roughly threefold. Under a random-pairing counterfactual derived from stated receptivity, only 4-13% of directed dyads combine agent deployment with receiver engagement, with a pronounced gender-directional imbalance. Design counterfactuals quantify the levers: a reciprocity requirement cuts interaction volume by half or more by excluding nearly two-thirds of would-be deployment, while routing agent contacts on receive receptivity triples per-contact engagement, a lift that survives out-of-sample validation with the target item held out (AUC 0.88, 3.1x quartile lift under respondent-level cross-validation). We discuss implications for agentic recommender design, including disclosure, opt-in mechanics, and receptivity-aware matchmaking.

Urban traffic congestion reduces productivity and increases travel cost and emissions. Network-wide live travel-time shortest-path rerouting can be highly effective in simulation, but assumes that essentially every on-road vehicle is replanned every decision period. We propose HLSR, a selective hybrid live--forecast vehicle rerouting framework that fuses live edge speeds with short-horizon forecasts under limited intervention scope. Building on dual-threshold congestion detection, calibrated upstream selection, and driver-tailored travel-time prediction, HLSR further introduces approaching-vehicle expansion, travel-time-weighted k-shortest-path generation, and a horizon-dependent hybrid live--forecast segment speed used in multi-cost route allocation.

Reliable quantitative analysis of dynamic contrast-enhanced MRI requires high-quality spatiotemporal reconstructions at high undersampling rates. Scan-specific reconstructions using Gaussian and Gabor primitives have shown promising results without the need for large training datasets, but have not addressed the additional dimension of dynamic contrast. We propose a multi-dimensional, primitive based framework for dynamic contrast-enhanced MRI reconstruction that disentangles the underlying anatomy, the dynamic contrast enhancement, and residual motion into separate temporal basis functions, thereby enabling a geometrical interpretation of the representation. We show that this architecture achieves performance competitive with conventional reconstruction methods, both in reconstruction quality and in the accuracy of extracted aorta and kidney enhancement curves. The modular tier design extends naturally to additional dynamic factors and higher acceleration rates. Code available at https://github.com/compai-lab/ 2026-GaborDCE-spieker.

AI agents increasingly perform knowledge work (i.e., produce and modify persistent digital artifacts such as code repositories, documents, spreadsheets, slides, reports), yet the parsed views they search, the native files they edit, the changes they review, and the artifacts they submit can refer to different versions of the same work product. We formulate this as a workspace-state contract: every view should be explicitly tied to a version of the evolving workspace state. Coding agents partly address this need through repository contracts for search, diffs, and tests, whereas an analogous contract is less explicit for PDFs, spreadsheets, slides, notebooks, and mixed-format project folders. We propose StagedWorkspace, a versioned workspace for knowledge-work agents. The workspace binds parsed records and review diffs to content hashes of the native files as they change. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points. SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, compared with published same-model scores of 29.3% and 25.5, respectively. A paired review-axis ablation on 57 file-editing tasks further finds higher observed scores when diffs are visible. These results identify workspace state as an experimental variable in knowledge-work agents and motivate benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.

Language has two parameters. Count how often words occur together and you estimate amplitude, the strength of association. Word embeddings and attention weights refine that count, which sums every writer in the corpus together. This paper claims a second parameter, phase, which signed weights learned from a corpus do not supply. Phase exists only between meanings: it determines how coactivated meanings combine, and it can reverse what a meaning contributes while that meaning stays fully present. A speaker can set phase in the signal through linguistic form; encounters install phase relations and history distributes them. Population averaging deletes history-indexed phase: agent-deindexed corpora identify the population marginal state and determine no individual or dyadic state, at any scale. The standard transformer has no explicit representation for phase in frozen inference, and the interpretability program measuring progress by monosemanticity is optimizing against it: the coexistence it treats as a defect is the condition of allusion, irony, and quotation. Six predictions test whether a suppressed meaning stays active, whether encounter order changes what a phrase does, whether marking the signal changes how a shared phrase is taken, and whether a model given a history is changed by it or only informed about it. The claim defended is the weak version: interpretation requires a second relational parameter, signed, persistent, and indexed to individuals and dyads. Quantum probability is one notation for the parameter; nothing in the formalism claims quantum processes in the brain. The strong version, that the quantum calculus constrains these phenomena as signed classical models do not, rests on an encounter-order constraint not yet derived. The architecture the theory calls for is a language model with agent-indexed, phase-bearing semantic states.

Sampling from a diffusion model typically requires many forward passes through a large neural network, making generation computationally expensive. While much work has focused on efficient solvers and samplers, comparatively little attention has been paid to selecting the sampling timesteps themselves. A recent line of work optimizes theoretically derived surrogates for sample quality rather than the quality metric itself. We propose Optimizing Your Sampling (OYS), which instead treats timestep selection as a black-box optimization problem, optimizing the target metric directly with Bayesian optimization. OYS outperforms both the default schedules and those of Align Your Steps on text-to-image generation, and improves over the default schedules on inpainting and other image tasks, in both quantitative and human evaluations. OYS requires no additional training, is applicable even to distilled models, and improves both simple and sophisticated samplers such as Euler and DPM-Solver++. A 5-step OYS schedule retains 89%-94% of the quality of a 50-step schedule while reducing inference cost by 10x.

MongoDB maintains client libraries in a dozen programming languages, used by tens of thousands of organizations and millions of developers. Most are implemented natively rather than as wrappers around a shared core. Ensuring consistent behavior across these libraries, comprising millions of lines of code, is hard but essential. Over eleven years, we developed a specification-based testing approach: tests are written once in YAML and executed by language-specific interpreters for each library. We describe the evolution from many ad-hoc formats to a Unified Test Format, which allowed us to delete over 22,000 lines of test code. The rate of nonconformance bugs fell up to 86% in drivers that adopted YAML tests (though results varied). We report lessons learned about declarative test design, test architecture, schema evolution, and the limits of unification.

MRI reconstruction methods for undersampled k-space data naturally utilize complex-valued measurements. Parallel developments in sparse phase retrieval have shown that magnitude-only measurements may provide complementary information for signal recovery. However, their use in MRI reconstruction remains largely unexplored, due to lack of practical settings where informative magnitude measurements can be obtained without additional scan time. In this work, we investigate the use of auxiliary k-space magnitude information for accelerated steady-state dynamic MRI reconstruction, and demonstrate strong consistency of k-space magnitudes across time-frames. Building on this observation, we propose $\mathbb{C}+\text{Mag}$, a magnitude-informed physics-driven deep learning reconstruction method. The proposed method employs an ADMM-based unrolling framework with a novel magnitude-aware data-fidelity formulation, where quadratically smoothed optimization and momentum-based updates are introduced to address the non-differentiability and non-convexity of the magnitude constraints. Experiments on retrospectively undersampled cine MRI and phase-contrast flow MRI datasets, as well as prospectively undersampled real-time cine MRI acquisitions, demonstrate improved artifact suppression, sharper anatomical recovery, and better preservation of phase information compared to conventional PD-DL methods, which is further supported through blinded expert reader evaluations.

Categorising invoices into the correct General Ledger (GL) code underpins financial reporting and tax compliance. This is a skilled accounting judgement rather than a routine task: the correct category depends subtly on the nature of the purchasing business, the vendor and the invoice text. Whilst AI is increasingly being adopted across industries to automate tasks, including invoice categorisation, implementations built on in-house small language models (SLMs) can simultaneously reduce cost and improve data security, confidentiality, and interpretability. We investigate this approach by first analysing the pre-trained embedding geometry of a small sentence transformer (SBERT) and classic SLM (DeBERTa). The sentence-embedding space of this financial corpus is globally anisotropic but composed of locally isotropic clusters, extending prior token-level findings to sentence embeddings in a financial setting, and these clusters are strongly correlated with the vendor identity. SBERT fine-tuned on a single GPU reaches 0.96 accuracy on invoice classification, above both a zero-shot LLM and a vendor identity baseline, increasing performance for smaller, challenging categories and new clients. For this important generalisation problem, SBERT reaches 0.9 F1 with roughly 100 client-specific invoices, showing that an in-house SLM implementation is promising. Combining these results with geometric analysis shows that pre-trained embedding geometry is associated with classification performance and reveals a counterintuitive finding that a structured input that would help a human reader does not improve the SLM performance.

Humans continuously learn from experience, whereas conventional large language model (LLM) evaluations ignore the models' ability to improve through inference-time interaction. In this paper, we study how LLMs learn from iterative experience at test time, a setting we refer to as Chain-of-Experience (CoE), where models accumulate experiential traces through iterative interactions with self or environmental feedback to form a continual improvement loop beyond zero-shot inference. We instantiate CoE with diverse feedback mechanisms, including model self-feedback and environmental signals such as correctness or public coding test pass rates, and evaluate across math, coding, and knowledge domains using 8 LLMs, including GPT-5, Gemini-2.5 Pro, Claude-4.5 Sonnet. Our study shows that leveraging iterative experience consistently outperforms feedback-free baselines, achieving substantial gains with self feedback alone, alongside a 5.6% overall improvement and 19% lower API cost across tasks and models. We further show that combining complementary feedback channels (e.g., model and correctness signals) yields additional gains, and that CoE delivers higher accuracy per token than existing test-time strategies. We observe a positive correlation between LLM base ability and improvement capacity, and show that models remain robust under weak or spurious feedback, with different feedback contributing to distinct improvement aspects and most gains emerging early in the iterations.

Large-scale, high-dimensional tabular regression remains challenging: tree-based models are robust but lack end-to-end representation learning, while deep models enable flexible feature learning but often incur costly interaction modeling and sensitivity to noisy or redundant features. We propose TabNSM, a scalable regression framework that extends our earlier sparse-attention and mixer architectures. At its core, the Adaptive Sparse Interaction Module (ASIM) integrates foreground feature discovery, sparse local interaction encoding, and Feature-Token Mixing, providing near-linear complexity under fixed sparse configurations. For regression, TabNSM introduces three complementary components: a Multi-Stage Regression Head for progressive prediction refinement; GridLoss, an ordinal-aware soft-binning objective that incorporates target structure into representation learning; and RISE (Reweighted Instance Sampling by Error), a difficulty-aware sampling strategy based on loss-quantile bins. Across nine real-world regression benchmarks, TabNSM delivers strong predictive performance and practical scalability, with particularly consistent gains on high-dimensional and heterogeneous datasets. These results demonstrate that selective interaction modeling, structured regression supervision, and difficulty-aware sampling provide an effective and scalable approach to deep tabular regression.

GPT-style models achieve strong performance by representing language with finite vocabularies of reusable discrete tokens. This success has motivated symbolic music tokenizations to treat recurring musical structures, such as chords, motifs, and phrases, as reusable units analogous to linguistic tokens. However, tokenization derives its advantage not from reusable combinations alone, but from compression: effective compression requires coordinates in which recurring regularities form stable and predictable conditional distributions. The key problem is therefore not to find larger musical combinations, but to discover the coordinate system in which musical facts become predictively compressible. We formulate the Effectiveness--Losslessness Framework and define tokenization as the construction of a predictively effective and relationally lossless coordinate system. The Predictive Effectiveness Principle defines the Fact--Token Boundary: decoupling and denesting construct coordinate interfaces that expose predictive regularities. The Relational Losslessness Principle defines the Token--State Boundary: tokenization stops before context-dependent relations are fixed, leaving their computation to model states. Controlled symbolic-music experiments validate these boundaries. Effective coordinate construction improves predictive compressibility, while fixed relational projections constrain contextual modeling. Sequence compaction alone does not guarantee predictive compression, while preserving contextual freedom allows higher-order musical organization to emerge without explicit structural labels. These results reveal why GPT-style models do not transfer directly across modalities: architectures transfer, but tokenization interfaces do not. Tokenization must discover effective representations while preserving the relational freedom from which contextual structure can emerge.

WEASEL 2.0 is a dictionary-based time series classifier that combines dilated sliding windows with a randomised hyperparameter ensemble and a fixed-size dense feature representation. Two of its hyperparameter choices, the maximum ensemble size and the maximum window size, are specified by simple thresholding rules whose chosen thresholds are not empirically justified in the original paper. In this work we reproduce WEASEL 2.0 on 114 UCR datasets, achieving a mean accuracy of 0.865 and median of 0.928, closely matching the published values (Wilcoxon signed-rank, p = 0.655). We then test the sensitivity of four design choices: the downstream classifier, the absence of feature weighting, the maximum window-size rule, and the maximum ensemble-size rule. The first three are robust to perturbation. The fourth is over-provisioned for long-series datasets, motivating an adaptive rule that sets the maximum ensemble size from series length and number of classes. Evaluated on fixed-length datasets, the adaptive rule reduces peak fit memory by a median of 37 MB (mean 395 MB) and fit time by a median of 0.4 s (mean 4 s), with a median accuracy change of 0% (mean -0.11%). Memory and time savings concentrate on long-series datasets where the original rule allocates the largest ensemble size.

Improving flight safety with flight data requires not only accurate detection of risk events, but more importantly, clear interpretation of their underlying causes at the level of pilot control behavior. Existing explainable AI techniques, such as feature importance maps, often require considerable domain knowledge to translate them into operationally meaningful explanations. Large Language Models (LLMs), which excel at language reasoning, bring a promising solution to this issue. However, applying LLMs in this domain presents key challenges such as modal inconsistency, limited classification ability, scarcity of task-specific data for fine-tuning, and lack of domain knowledge. To overcome these challenges, we propose FlightLLM, a prior-guided semantic LLM-based approach for interpretable flight safety analysis. Specifically, we first perform feature engineering to address modal inconsistency, combining statistical descriptors with physically meaningful flight indicators. This representation is further processed by a Semantic Discretization module, which converts abstract numerical patterns into qualitative descriptions that are more compatible with language reasoning. In addition, since LLMs are not inherently strong classifiers, CatBoost is incorporated as a statistical expert, and its prediction results are injected into the prompt as prior guidance. A contrastive few-shot learning strategy is further adopted to compensate for limited data. Finally, we design structured prompts to embed aviation-specific knowledge into the inference process. Using hard landing, a representative risk event with complex causal mechanisms, as an anchor point, we evaluate FlightLLM on a dataset of 704 real-world A320 flight samples. Experimental results show that the proposed approach achieves competitive classification performance while generating direct and reasonable explanations for event causes.

Reasoning in LLMs is overwhelmingly studied in domains that provide a model with rules: mathematics and code. Linguistic puzzles invert this: the solver must first discover the system before reasoning within it. We present the IOL-AI Challenge, an open-science competition run on the unseen problems of the International Linguistics Olympiad (IOL) 2026 Individual Contest, evaluated both automatically and, for the first time, by members of the official IOL Jury under the same rubrics applied to human contestants. The challenge drew 731 submissions from 46 teams under a strict compute budget (one T4, 30 mins). We additionally benchmark 15 unconstrained frontier and open models, with Claude Opus 4.8 earning a jury score equivalent to a gold medal, while both resource-constrained systems we submitted for jury grading scored in the range of the bottom 5% of contestants. Capability was not determined by scale: 14B submissions outperform models twice their size, and gains come from decoding and output-handling rather than model capacity. We also found that automatic metrics rank systems exactly as the jury does, but compress the scale, upscoring weak systems by ~13 points and understating strong ones. Our analysis shows that while frontier models might have prior knowledge about some of the problem languages, it does not significantly help them solve the linguistic reasoning tasks, leaving linguistic reasoning as a strong benchmarking proxy for generalizable reasoning skills.

Combining large language models with reinforcement learning is increasingly explored, yet the theoretical status of LLM-derived reward signals is often left implicit. We formalize the hybrid LLM-planner and RL-controller architecture as a Goal-Augmented Markov Decision Process and show that when the LLM per-state progress score is used as a bounded potential function, the resulting shaping term preserves the optimal policy set even when the LLM scores are inaccurate. This guarantee is stronger than what general LLM-as-reward approaches provide. We verify the result numerically on a small MDP under four potential configurations, including an adversarial one scaled to twenty times the base reward magnitude.

Probabilistic modeling of physical fields benefits from both a data-driven prior and known physical structure such as the governing equations. Energy-based models (EBMs) are a natural fit since energies compose additively, which enables augmenting physics information during inference. However, EBMs have been difficult to train and sample from due to the intractable partition function. We show in this work that flow matching models with a potential-induced velocity yield an explicit scalar energy at all transport times, whose gradient is exactly the converted learned score and which recovers the marginal negative log-density at the population optimum. The time-dependent energy functions are obtained purely from the matching regression objective on an independent linear Gaussian interpolation, without a variational form or additional MCMC steps, and the sampling retains the flow ODE. Access to the energy function from a trained model serves three roles: energy-corrected data generation, energy as a scoring function for out-of-distribution (OOD) detection, and energy compositional posterior sampling for inverse problems. In particular, we show the explicit energy permits general MCMC samplers in the predictor-corrector sampling framework, reducing PDE residual and spectral distance compared to the flow ODE baseline. Furthermore, we demonstrate utilizing the data energy and physics-based energy (e.g., PDE residuals) as complementary mechanisms to improve detection accuracy for OOD tasks. In addition, we explore the connection to MCMC-based inference for inverse problems by composing the energy with a quadratic observational likelihood that yields a posterior energy, used as an explicitly chosen family of inference-time targets.

Artificial intelligence (AI) is becoming part of the working infrastructure of the biosciences. AI models can predict biomolecular structures, design proteins, rank variants, annotate images, recommend strains and optimise experimental conditions. We argue that the decision to use an AI output to guide laboratory action is a key juncture for trustworthy research and should follow a defined, reviewable process. We propose Traceable Trust as a proportionate assessment-and-design framework for this output-to-action boundary. It asks what evidence supports the output, what capability is being claimed, what agency has been delegated, what threshold authorises action, who can override it and how outcomes inform later decisions. We illustrate the framework through three case studies spanning ecosystem resources, project design and laboratory action. Together, the cases show how trust can be documented where AI outputs begin to shape scientific work.

Using LLMs as judges has become standard practice for evaluating model outputs at scale. This is particularly common for subjective, open-ended tasks such as assessing helpfulness or alignment, where no single reference answer exists. However, objective tasks introduce a distinct reliability challenge for reference-free LLM judging. In the absence of a reference answer, the judge evaluates factual correctness either through its parametric knowledge or through tool augmentation. Although the former enables efficient evaluation, the judge may hallucinate or lack sufficient evidence for its verdict. Conversely, tool augmentation can provide additional evidence but introduces extra computational cost and requires an appropriate mechanism to determine when and how that evidence should be used reliably. More importantly, neither approach alone provides formal control over the risk of accepted verdicts or guarantees their reliability at a specified level. We propose a risk-controlled framework that calibrates uncertainty thresholds on a held-out set so that the false discovery rate among accepted verdicts remains below a user-specified level~$α$ with high probability, using finite-sample Clopper--Pearson intervals. When the parametric mode is not sufficiently confident, the instance is routed to a retrieval-augmented mode, where the judge gathers web evidence and re-evaluates the instance under a second calibrated threshold. The finite-sample guarantee carries over to this two-threshold routing without additional assumptions. Across open-domain QA benchmarks and judges of varying scales, the framework maintains the target error rate while achieving substantially higher coverage than single-mode baselines.

This paper reports on a locally motivated mixed-methods case study addressing a teaching-related suspicion held by software engineering instructors: that commit messages written by students frequently fail to serve their intended communicative role. To examine this suspicion empirically, we analyzed commit messages from student and industrial case-study projects using a partial replication of an established commit-message quality taxonomy. The results confirm that communication and quality issues occur recurrently in both contexts, substantiating the instructors' initial concern. Motivated by this finding, we devised What Do You Mean? (WDYM), a lightweight, role-based educational game intended to surface and address commit-message communication breakdowns within the constraints of university coursework. Analysis of gameplay observations and participant surveys shows that WDYM is effective in raising awareness and fostering reflection on commit-message communication issues, although it provides limited evidence of immediate or sustained improvement in commit-message writing practices. Taken together, the study presents WDYM as a useful, though imperfect, context-bound intervention for addressing a locally observed instructional problem, without claiming general applicability beyond the studied setting.

The rapid growth of social media has greatly influenced political discourse, highlighting the need to understand individual political ideologies and their temporal dynamics. This task faces challenges such as data scarcity, abundant non-political content, costly and bias-prone manual annotation, and difficulty in modeling future ideological inclinations. To address these issues, we propose TSN4PI, a unified framework for tracking the evolution of political ideologies on social media. It includes two core modules. The PIDN uses large language models with style transfer and unsupervised domain adaptation to enable robust ideology detection and filter irrelevant content from noisy, cross-domain data. The PIPN employs temporal graph neural networks to predict future ideological shifts, enabling comprehensive analysis of ideology presence, intensity, and evolution. We release two large-scale datasets for noncommercial research use to facilitate further work. Extensive case studies on multiple platforms (X and Truth Social) validate the effectiveness of TSN4PI and provide empirical insights into political polarization and the evolution of online ideologies. Our findings offer a nuanced perspective, advancing both methodological development and empirical understanding in this field.

Ultrasound tongue contour segmentation remains challenging under cross-dataset domain shift, where limited annotations, probe variability, and acquisition noise often degrade model generalization. We present a source-free domain adaptation framework for robust ultrasound tongue segmentation built on a lightweight UltraUNet backbone. Starting from a checkpoint pretrained on only five labeled source images, simulating an underfitted constrained source model, the proposed method adapts to a fully-unlabeled target domain by iteratively refining pseudo-labels, filtering unreliable masks with a contour-based quality-control module, and generating target-style synthetic image-mask pairs through a segmentation-guided conditional GAN. The student model is then trained on a mixture of clean pseudo-labeled target images, noisy pseudo-labels with consistency regularization, and synthetic samples, enabling closed-loop adaptation without access to source data. We evaluate the method on 12 source-target transfer pairs across eight ultrasound tongue imaging datasets, and conduct source-size scaling experiments and ablation studies. Across all comparisons, the proposed framework improves segmentation overlap and contour accuracy over the baselines, including supervised ones. These results suggest that task-specific pseudo-label refinement and synthetic target-style augmentation can substantially improve source-free adaptation for ultrasound tongue imaging.

We describe an inference-time architectural enhancement for off-the-shelf foundation models that markedly reduces perplexity and boosts accuracy across generation and reasoning tasks. Our approach incurs essentially no additional latency during generation, though it requires serial processing in the prefill phase. Motivated by the fundamental limitation that state updates in feedforward transformers are bounded by model depth, our technique, recirculation, introduces a specific form of recurrence that allows the model to act as a dynamical system and track belief states. We distinguish this technique from chain-of-thought computation---which is better reserved for complex inferences rather than basic state tracking---as well as from popular depth-recurrence techniques (looping) and the costly training of recurrent transformers. We also propose and evaluate an adaptive variant of recirculation which requires only light tuning of hyperparameters while freezing the original model weights. Relative to the off-the-shelf baseline, adaptive recirculation achieves remarkable gains on the Gemma3 family, including a 23% reduction in perplexity on a suite of datasets, a 21% increase in accuracy on GSM8k, and reliable improvements in accuracy on other downstream tasks. Our training-free approach succeeds by leveraging the model itself to inform architectural modifications, suggesting a route to architectural evolution guided by a trained network's properties rather than forced, arbitrary design choices.

Authorship verification (AV) assumes that an author's writing style remains sufficiently stable to distinguish it from that of other writers. In practice, however, this assumption is challenged by distribution shifts caused by changes in genre, time, and AI-assisted writing. Existing AV benchmarks typically study these factors in isolation and focus predominantly on English, limiting our understanding of model robustness under realistic conditions. We introduce AVShift, the first German benchmark for systematically evaluating AV under multiple distribution shifts. AVShift comprises over 150,000 text pairs spanning three genres and 21 years, enabling controlled evaluation of cross-genre, temporal, and AI-era shifts within a unified framework. We benchmark representative feature-based, embedding-based, and LLM-based approaches. Our experiments show that fine-tuned LLMs generalize best across genres and benefit substantially from stylistically diverse training data. We further demonstrate that temporal drift is one of the strongest factors affecting AV, with performance degrading significantly as the time gap between documents increases. In contrast, we find no evidence of a measurable AI-era distribution shift within AVShift. Finally, our feature analysis reveals stylistic features that remain stable across genres, while their relative importance varies depending on the specific genre transition. We release AVShift and our code for future research.

The impact of climate variability on food production has led to the creation of various forecasting models that uses machine learning (ML), numerical weather predictors (NWP) or a hybrid of ML-NWP models to identify structural and physical relationships between meteorological drivers and crop growth, in order to predict crop yield. Droughts, for example the 2012 Midwestern US (Corn Belt) drought, are extreme events that affect crop production and test the limits of these forecasting models. Using 16 meteorological drivers as predictors, we compare ML (non-deep learning) and deep learning forecasting models to predict the county-level corn yield for the extreme drought year, 2012. This forecasting problem is characterized by a dissimilarity between the feature distributions of the training and test data, where the meteorological conditions of the extreme drought year fall outside the range of historically observed values. Additionally, the dataset consists of spatial and temporal irregularities where counties with missing yields introduce spatial sparsity and the use of only a subset of daily values per year introduce temporal sparsity. To overcome this, we use sample weighting and feature selection as modifications to improve our forecasting models. These modifications lead to an improvement for ML models; however, the deep learning model VITA shows little to no improvement. While VITA outperforms the ML models with or without modifications, our current study sheds light on the effect of dissimilarity between train and test feature distributions on forecasting models, compares deep learning versus non-deep learning models, and introduces modifications that are effective for non-deep learning models.

Online log anomaly detection is critical for maintaining the reliability of large-scale computing systems. Although recent language model-based log anomaly detectors achieve strong detection performance, their confidence estimates remain poorly calibrated. We show that these detectors frequently assign excessive confidence to incorrect predictions, particularly for anomalous logs under severe class imbalance. Moreover, confidence on erroneous predictions remains persistently high even when conventional calibration metrics indicate good calibration, creating a critical reliability gap for operational monitoring systems. To address this issue, we propose Log Reconstruction and Distance (LoRD), a lightweight post-hoc calibration framework for reliable log anomaly detection. LoRD learns prediction-route-specific reliability models from latent representations of correctly classified validation samples and estimates prediction reliability through route-wise reconstruction distances. Based on the estimated reliability, LoRD selectively recalibrates high-risk predictions to suppress overconfident errors while preserving reliable predictions. Extensive experiments on four large-scale log benchmark datasets and multiple language model-based detectors demonstrate that LoRD consistently improves confidence reliability and substantially reduces overconfident anomaly-related errors without sacrificing anomaly detection performance.

State-of-the-art model-based reinforcement learning methods learn neural world models that allow policy improvement by planning in a latent space, without assumptions on the structure of the underlying environment. While expressive, these models are generally task-dependent: they learn uninterpretable latent representations that are tied to the training task and thus hard to generalize to new tasks. In this work, we present a novel world model formulation where the reward prediction only depends on a subset of structured, symbolic components of the whole latent state. Decoupling observation reconstruction and reward prediction allows us to learn world models that can adapt zero-shot, i.e. without further environment interactions, to new reward functions defined over the same symbolic state space. We discuss the main advantages and challenges of learning these neurosymbolic world models and demonstrate the strong generalisation properties of our approach over purely neural methods.

Tabular Foundation Models (TFMs) increasingly rely on in-context learning, where a model receives labelled examples at inference time and predicts labels for new inputs without updating its weights. Existing TFMs are typically trained on either massive synthetic corpora or very large collections of real datasets. In contrast, we show that surprisingly strong transfer can emerge from self-supervised pre-training on just a single real table. In this setting, we also find that tables tend to be either broadly useful or broadly poor regardless of downstream prediction task, and that the strongest predictor of usefulness is the number of features rather than the number of instances. This leads to a task-centric interpretation of tabular pre-training: the number and the quality of tasks are essential for the pre-training of TFMs. We show that the same task-centric perspective can help corpus design at scale: fine-grained column-level pre-processing consistently improves downstream performance, while no improvements are observed when we filter or deduplicate at the dataset level. Finally, we offer a new perspective for how TFMs generalize: we believe that tabular in-context generalization is largely retrieval-based, and good models are those that learn to identify relevant examples in the provided context and aggregate them well. The mechanics of TFMs have been relatively understudied; our task-centric, retrieval-based perspective offers a new framework to guide future model and corpus design.

In the Code World Model paradigm an LLM synthesizes an executable world model that a classical planner searches, and the model is accepted when it reproduces sampled transitions. We ask what that acceptance certifies in continuous control. We define the pipeline's danger as an expected risk and isolate its exact factor: the probability that N i.i.d. gate rollouts all miss a critical event of probability r is exactly (1-r)^N; an independent acceptance sample adds its budget to the exponent. On three hybrid instruments the accepted mode-blind model is exploited: the planner is pinned at the mode boundary at a regret of nearly the whole attainable return. We prove a localization budget, valid at boundary points: models with Lipschitz constant at most L differing by eta at a point disagree above tolerance eps on a region of volume at least kappa((eta-eps)/L)^(d+m); the discontinuous reset modes studied pay no such budget. With real LLM synthesis, GPT-5.x repairs an omitted 1D clamp in 105 of 111 mode-containing draws -- every attempt exact on 50 of 56 instrument-stream blocks (95% CI [0.781, 0.960]). On 2D regions no artifact recovers the rule (0/156); eight targeted interventions leave the failure in place, and positive controls locate it: a located rule is not induced, while given form and location the constants follow exactly. A version-space certificate proves identification is class-relative: at the widest dose the declared fit succeeds in 20/20 blocks and every sample-consistent circle is within tolerance in 18/20. We prove a class of entry rules exactly consistent with every sample yet harmless at play, so identifiability is a measurable property of the instrument. Re-scoring all 1034 artifacts on independent samples confirms acceptance certifies sample consistency and no more: where the gate is provably informative it covers about two percent of the exploited planner's queries.

Large Language Models (LLMs) demonstrate remarkable multi-hop reasoning capabilities over long contexts, yet the internal mechanisms enabling these distant cognitive leaps remain poorly understood. Traditional attention-based interpretability often fails to capture true semantic proximity due to routing artifacts like attention sinks. In this paper, we bypass attention weights to directly analyze the dynamic geometry of the hidden state manifold, proving that deep LLM latent spaces natively organize into Small-World networks. By sparsifying the continuous similarity matrices of long-context representations into unweighted graphs, we trace the connectivity between highly disjoint semantic anchors across two distinct architectures. Our findings reveal a sharp topological phase transition: while early syntactic layers remain entirely fractured, deep reasoning layers abruptly compress massive conceptual distances into highly navigable pathways strictly bounded by the "Six Degrees of Separation" limit (=< 6 semantic hops). Furthermore, we demonstrate the practical efficacy of this framework by applying it to zero-shot hallucination detection within Retrieval-Augmented Generation (RAG) using the RAGognize dataset. We show that factually grounded generations maintain structural integrity with their source context (approximately 3 hops), whereas hallucinations induce severe topological collapse. Ultimately, this work mathematically formalizes how transformers execute abstract reasoning and provides a novel, strictly geometric signature for evaluating factual reliability.

Recent research has leveraged Large Language Models (LLMs) to enhance Automated Feature Engineering (AutoFE) through semantic descriptions and trajectory-based prompting. However, there exist two challenges that limit their applicability and scalability in long-horizon optimization: (1) semantic metadata is unavailable in many practical settings, and (2) trajectory accumulation increases the risk of exceeding the context window, while without it, the generation process can become unstable, leading to becoming stuck in the local optima and a high duplicate rate of generated features. To this end, we propose a SHAP-enhanced Implicit-trajectory Generation for Metadata-free AutoFE (SIGMA), a scalable constant-context optimization framework. SIGMA leverages SHAP values to provide task-aware signals for guiding group feature generation instead of semantic information. In addition, we adopt an EXposed-feature Implicit Trajectory (EXIT) approach, where the exposed features in the prompt implicitly represent the trajectory. Empirical results demonstrate that SIGMA achieves performance comparable to the state-of-the-art (SOTA) LLM baselines with a nearly constant prompt length. Notably, EXIT significantly reduces the duplicate ratio of generated features from 37.2% to 6.8%. At the same time, SIGMA matches traditional SOTA performance with only 5.4 features on average, demonstrating substantial efficiency gains in feature utilization.

Large language models can generate executable programs, which makes it possible to search directly over procedural content generators rather than individual levels. We study this approach in Sokoban, Zelda, Dangerous Dave, and Lode Runner. Each run evolves complete Python generators through language-model mutation and crossover. We introduce Continual Abstraction Discovery, or CAD, which extracts reusable primitives from high-fitness programs into a run-specific helper module. A 2x2 experiment crosses CAD with access to a fixed hand-written domain API. The completed data set contains 160 complete runs, with at least ten 50-generation runs in every cell. CAD raises mean final best fitness in all eight domain and API comparisons. Across all CAD runs, learned libraries are adopted by most later programs and repeatedly rediscover validation, reachability, and structural utilities. These results support that discovering reusable primitives improves evolutionary program search for content generators.

Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models but relies on costly rollout exploration. Assigning the same exploration budget to samples with different difficulty levels is inefficient: easy samples may receive redundant rollouts, whereas difficult but learnable samples may receive too little exploration. Existing adaptive schedulers address this mismatch through curriculum-based sample selection or non-uniform rollout allocation based on estimated sample difficulty. However, obtaining reliable online difficulty estimates remains challenging: dedicated probing adds substantial generation overhead, whereas history-based estimators face a cold start with no initial observations and stale feedback, and typically ignore relations among samples. To address these limitations, we propose a plug-and-play graph-based online difficulty estimator that shares rollout feedback across related samples and continuously updates their difficulty estimates, mitigating cold start and staleness without dedicated probing. Specifically, we first construct a difficulty-aware sample graph based on semantic and reasoning similarities. Based on this graph, we introduce latent difficulty states and use a Potts prior to encourage neighboring samples to share the same state. We then employ a state-level Beta-Binomial model to aggregate the rollout outcomes associated with each state. Finally, we use an online mean-field variational algorithm to continuously update the latent-state assignments and state-level difficulty as new feedback arrives. Our framework can be integrated into sample-selection and rollout-allocation schedulers, enabling difficulty-adaptive exploration without dedicated probing. Experiments across multiple base models, RL schedulers, and benchmarks demonstrate that our framework achieves better performance.

Small language models can grade open-ended examination answers as reliably as substantially more expensive models when they grade against an explicit rubric. We test this claim as the design principle behind any-to-bench: a frontier model reads source documents once, at ingestion, to extract each question and its rubric; lower-cost models then perform all repeated grading work. We evaluate six cost-efficient model configurations from two model families at three reasoning-effort levels. Each configuration answers 24 open-ended examination questions, and each also grades every answer sheet three times, yielding 3,456 per-question grades. Scores depend overwhelmingly on the answer being graded: answer identity explains 95.6% of score variance, whereas judge identity explains only 0.2%. Raising a writer's reasoning effort moves earned scores by as much as 0.143 of full marks, while raising a judge's reasoning effort moves assigned scores by at most 0.006. Six frontier-tier judges, added as a check, reproduce these scores and are no more reliable as a panel. Two ablations then decompose the rubric on the same questions and answers. Removing its criteria and levels while keeping the official answer changes nothing measurable. Removing the official answer as well collapses reliability (ICC 0.888 to 0.628), inflates scores, and makes judge reasoning effort matter again. The rubric is what decouples grading from judge intelligence, and within the rubric the official answer does nearly all the work. We find no evidence of length preference or same-family preference under rubric-anchored grading.

Financial time series exhibit non-stationary and heterogeneous statistical properties, making change-point detection challenging because no single unsupervised algorithm performs consistently across assets and market regimes. Conventional workflows consequently depend heavily on expert-driven model selection, feature design, and hyperparameter tuning, limiting their scalability and adaptability. We propose EvoTS-Agent, a validation-guided self-evolving LLM agent for autonomous financial time-series change-point detection. EvoTS-Agent first performs curated exploratory data analysis to characterize dataset properties and initialize candidate detection models. It then evolves executable experiment trajectories through three complementary operators: \textit{Revision} exploits the current best solution, \textit{Alternative Strategy} explores fundamentally different modeling directions when progress stagnates, and \textit{Recombination} synthesizes complementary evidence from high-performing trajectories. Validation feedback guides trajectory evolution throughout the search, enabling the agent to adapt its detection pipeline to the statistical characteristics of each dataset while preserving reliable optimization. Experiments across four benchmark datasets demonstrate that EvoTS-Agent consistently outperforms existing LLM-based agents while maintaining a 100\% execution success rate across all evaluated backbone LLMs.

Groups routinely complete projects that no single member can plan, execute, or verify alone. We propose a formal model of this phenomenon, Collective Counterfactual Planning (CCP), in which the binding limitation on each agent is neither capability, knowledge, nor observability, but representational geometry: each agent perceives the state, conceives moves, consents to actions, and certifies goal requirements only through a projection onto an agent-specific subspace of a common task space. Four gates jointly determine whether a team can reach a conjunctive goal and legitimately recognize that it has done so: the exogenous implementation coalitions required to perform each action, together with three representational gates -- conception, consent, and task-relative verification qualification. We define the Collective Counterfactual Solvability (CCS) problem, separating geometric feasibility, executable attainment, and validated completion. The results expose a positive-negative duality. Iterated cross-agent relay can unlock a solution that no one-shot pooling of individual plans contains, but any goal requirement depending essentially on the subspace dark to the entire team is unverifiable and therefore not validly completable, even when the trajectory accidentally attains it. Memoryless and audited consent further constrain different objects -- action directions versus cumulative trajectory states -- and neither dominates the other. A four-step exhaustive horizon-bounded solvability scheme is sound and complete under exact representation of the relay closure; restricted implementations remain sound on returned plans but need not be complete. The model gives one geometry for sequential mutual enabling, competent execution of steps whose purpose is invisible to the executor, forced sub-teaming at expertise boundaries, and completion that cannot be validly declared.

Recent advances in AI have revolutionized speech processing, yet effective speech understanding requires discerning not just what is said, but how it is said. Speech Sentiment Analysis plays a critical role in decoding these paralinguistic cues for diverse real-world applications such as recruitment and customer service. However, existing Speech Sentiment Analysis research faces two primary limitations. First, dominant approaches rely on text-centric pipelines that cascade Automatic Speech Recognition with text analysis. This process inevitably discards essential acoustic features like prosody and tone, failing to capture attitudinal meanings in acoustically ambiguous utterances. Second, current benchmarks suffer from a mismatch in label granularity, prioritizing basic emotions (e.g., happy, sad) over the nuanced interpersonal stances (e.g., confident, impatient) necessary for social sensitivity. To address these limitations, we propose a novel dataset, SpeechSense, for fine-grained speech sentiment analysis. Specifically, we define a specialized 8-class taxonomy of interpersonal stances detectable primarily through prosodic cues beyond lexical content alone. We then construct a curated dataset based on this taxonomy, built from high-fidelity speech synthesis and rigorous human validation. Comprehensive experiments across multi-modal LLMs, text-only LLMs, and speech encoders demonstrate that models with acoustic access consistently outperform text-only baselines. These results empirically validate the primacy of acoustic cues in detecting subtle speaker attitudes, highlighting the necessity of SpeechSense. Dataset and supplementary materials are available at https://github.com/Sher13cked/SpeechSense.

Robust Markov decision processes optimize one policy against a set of plausible transition functions. This can be conservative when the unknown dynamics are fixed and become partially identifiable after deployment. We study adaptive policy portfolios: finite sets of memoryless randomized policies synthesized offline and paired with a lightweight online selector. Robust regret is a natural measure of portfolio quality: for each plausible environment, it measures the loss of the best portfolio member relative to the policy that would have been optimal had that environment been known. Related regret objectives were studied by Ghavamzadeh et al. (2016) with an emphasis on approximations and relaxations for safe policy improvement. We give a complexity-theoretic account of portfolio certification and synthesis. Certifying a given portfolio is $\forall\mathbb{R}$-complete already for deterministic portfolios in acyclic (s,a)-rectangular RMDPs. Synthesizing a portfolio of unary-bounded size is $\exists\forall\mathbb{R}$-complete for general rational polytopes, even with fixed discount and acyclic dynamics. The single-policy case is already hard, both combinatorially and algebraically. Finally, we present an offline portfolio construction that is amenable to runtime specialization.

In the Lifelong Multi-Agent Path Finding (L-MAPF) problem, agents must repeatedly move from one destination to another while avoiding obstacles and inter-agent collisions. Widely regarded as one of the highest-performing solutions to this problem is the Rolling-Horizon Collision Resolution (RHCR) framework. However, commensurate with its quality solutions, it incurs a computational cost that limits its applicability to even modest agent counts. In this paper, leveraging theoretical methods from the Locally Interdependent Multi-Agent MDP literature, we first theoretically prove the near-optimality of RHCR in a discounted MDP formulation of the L-MAPF problem. Then, we leverage these results to naturally motivate an extended framework called Group Decentralized RHCR (GD-RHCR) which incorporates a group decentralized structure that partitions agents based on a transitive communication scheme and plans for each partition of agents in parallel. We show that both RHCR and GD-RHCR achieve similar exponentially close to optimal guarantees, establishing a theoretical duality between the time based restrictions performed by vanilla RHCR and the additional space based partitioning performed by GD-RHCR. Lastly, we show that across varying maps, GD-RHCR is able to attain high throughput that scales into higher agent counts while maintaining a significantly lower per plan cost.

Appendicitis is one of the most common abdominal emergencies worldwide and requires prompt diagnosis and treatment to prevent life-threatening conditions. However, accurately differentiating complicated cases, such as perforation or abscess formation, from uncomplicated appendicitis remains a significant clinical challenge. Among other methods, ultrasound is a safer and more cost-efficient diagnostic technique because of the lack of radiation exposure. In this research, an advanced system capable of automatically detecting complicated appendicitis from ultrasound images was developed. A dataset consisting of 4679 ultrasound images with 5 classes, namely perforated, abscess, acute, appendicolith, and normal, was used for the proposed model training and testing. Four pretrained deep learning models, DenseNet201, InceptionV3, ConvNextTiny, and VGG19, have been employed for detecting and classifying complicated appendicitis. In the initial configuration, InceptionV3 achieved the second highest accuracy, with a value of 69.21%. Owing to suboptimal performance with raw images, further optimization techniques, including image preprocessing, hyperparameter tuning, model fine-tuning, and image sharpening, were applied. These enhancements significantly improved the model's performance, with an accuracy of 95.58% for InceptionV3. The model performance is then explained with gradient-weighted class activation mapping (Grad-CAM), which creates a heatmap of the regions responsible for the model's prediction of the infected areas. This could make crosschecking with experts much easier.

Background and Context: Question and inquiry are integral parts of knowledge seeking and learning. Despite their importance, students tend not to ask enough questions in the classroom. However, studies have shown that students interact extensively with generative AI systems for learning and problem solving. Objective: In this paper, we seek to better understand the types of questions that students ask AI systems, and how those questions evolve during problem solving and across tasks. Method: We use the Graesser et al. taxonomy to classify students' inquiries into 18 types. We develop a few-shot learning approach to automatically classify students' interactions with AI into these categories. We use this system to analyze 830 interactions of CS2 students across two programming tasks. Findings: Our results suggest that a small subset of question types accounts for the majority of student inquiries, and that the types of questions students ask change substantially as the task progresses.

Automatic Target Detection and Recognition (ATD/R) is critical for military decision support and (semi-)autonomous operations. Recent advances in object detection and artificial intelligence (AI) significantly boosted the potential performance of ATD/R. However, the scarcity of publicly available military datasets limits the application of these systems. As a solution, this paper explores the use of publicly available models and civilian datasets to achieve reasonable performance in military contexts. We benchmark several state-of-the-art models, including six iterations of the YOLO series and two variations on the DETR framework, on a newly acquired military relevant dataset. This dataset features military vehicles and challenging circumstances, including various degrees of occlusions and small targets. The out-of-the-box version of each model is validated alongside a version finetuned on the VisDrone dataset. This dataset features small objects, an Air-to-Ground (A2G) perspective and relevant classes, potentially generalizing to our military ATD/R task. We compare the performance of the models using mAP@0.5 and mAP@0.5:0.95, across A2G and Ground-to-Ground (G2G) perspective, target size and model size, giving insight into the real-time capabilities of models. Our main findings are: (1) bigger models outperform smaller models, (2) DETR-based models show promising results compared to the YOLO series,(3) fine-tuning models on an out-of-domain A2G dataset, improves their A2G performance and slightly improves their performance on small objects, but (4) all models still struggle with detecting small objects in an A2G scenario. We conclude that, despite recent advances in object detection, in-domain training is still crucial for creating capable ATD/R systems.

Static Timing Analysis (STA) is a critical step in the design flow of digital integrated circuits, however, obtaining accurate delay estimations can represent a challenge when limited information regarding physical design is available. In response, this work presents a hybrid and light-weight machine learning (ML) based approach that combines a decision tree with linear regression to improve pre-routing delay estimations generated by the open-source RTL-to-GDSII tool OpenLane. The proposed model achieves an 80\% reduction in error compared to OpenLane's estimates, demonstrates a 71\% improvement even without utilizing OpenLane-specific parameters. Overall, this method offers an alternative to traditional delay propagation techniques and more complex machine learning models that is not only accurate, but is also over 300 times smaller, 2 times faster and offers a higher explainability.

As LLM agents operate across structured workflows and sessions, preserving long-term history does not ensure that later contexts can recover relevant evidence through a bounded memory interface. We study this evidence-reachability problem in long-term conversational memory, where retrieval still relies heavily on semantic similarity. This works well for topical recall, but it often misses earlier experiences, plans, or motivations that are semantically distant from the later events they help explain. Existing memory graphs provide cross-memory structure, yet links driven mainly by semantic overlap can duplicate what the host retriever already recovers. We argue that link construction should instead prioritize a sparse set of retriever-complementary associations. We present CABLE (Complementary Antecedent-Based Linking and Expansion), a plug-in augmentation that constructs links designed to extend the host retriever's direct semantic reach. For each new memory, CABLE generates antecedent-oriented queries, retrieves prior memories, subtracts candidates in the direct semantic neighborhood, and verifies the remainder before adding the accepted complementary associations into a sparse directed graph. At retrieval time, CABLE expands the host system's retrieved seeds along these links to surface implicit supporting evidence. We evaluate CABLE with A-MEM on LoCoMo and MA-LongMemEval, and further integrate it into SimpleMem and Mem0g on LoCoMo, using Qwen3.5-27B, DeepSeek-chat, and GPT-4o-mini. CABLE yields higher mean LLM-judge scores in every evaluated system-level setting, with the largest gains in categories where useful evidence is distributed across memories or sessions, including open-domain, multi-session, and preference-oriented questions. These results support prioritizing sparse, reasoning-relevant associations that complement rather than duplicate the host retriever.

Autonomous research systems are increasingly capable of executing long research workflows, yet automation alone does not ensure that the resulting process remains scientifically grounded. We introduce AutoResearch, a two-stage system that connects Idea Generation with Idea Execution to address both how research ideas are formed and how they are reliably established through experimentation. In Idea Generation, AutoResearch continuously integrates emerging research signals with accumulated domain knowledge, identifies transferable mechanistic insights, and uses multi-model generation and cross-review to produce grounded, testable research plans. In Idea Execution, coordinated agents decompose these plans into experiments, iteratively implement and diagnose them, and employ independent evidence-based review before accepting research conclusions. Across representative settings in cross-modal retrieval, systems optimization, and benchmark-driven machine learning, AutoResearch turns generated ideas into measurable progress, detects and corrects unreliable experimental results, and makes evidence-conditioned decisions to continue, revise, or terminate research directions. For example, on RSICD benchmark, an AutoResearch-generated idea improves mean Recall from 32.84 to 34.69, while recording only 5 audit-confirmed issue events compared with 11-27 for other autonomous research systems. These results demonstrate a research process in which meaningful insight is grounded before experimentation and conclusions are grounded before acceptance: Insight In, Hallucination Out.

Recurrent models process long contexts efficiently by compressing their history into a fixed-size state, but modern architectures typically do so in a single causal pass over the sequence. Each input must therefore be compressed before the model knows how it will later be used, forcing a limited state to compromise across possible future demands. We introduce dynamic compression, which allows a recurrent model to selectively revisit past tokens and revise its fixed-size state through additional recurrent updates. The model need not preserve every part of the history at uniformly high fidelity in its recurrent state, because lower-fidelity information can be revisited from the retained raw sequence when it becomes relevant. We study this in a controlled setting where the model first learns multiple functions in-context and, later in the same sequence, encounters a series of few-shot tasks that each require it to identify and reuse one of those functions. A single-pass model must preserve every function at sufficient fidelity for any future task, whereas selective re-scanning allows the model to revisit and refine only the function currently needed. We find that dynamic compression substantially reduces the recurrent state required for accurate reuse and scales more favorably as the number of stored functions grows. These results demonstrate a computation--memory tradeoff in which recurrent models can spend more computation revisiting their history to make more effective use of a fixed-size state.

While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains incompletely evaluated. Existing benchmarks emphasize information extraction, require external domain knowledge, or cover professional documents only as one of many settings. They are also largely English- or Chinese-centric, leaving other languages and Russian, in particular, substantially underrepresented. To address these limitations, we introduce BEAR-Bench (Bilingual Enterprise and Academic Reasoning), a self-contained, complex English-and-Russian benchmark comprising 1000 human-annotated questions based on text-rich business and scientific documents. We evaluate 16 proprietary and open-weight MLLMs, including Gemini 3.1 Pro and Qwen3.5-397B, on BEAR-Bench and observe clear headroom even for the strongest systems. Finally, we use the resulting model outputs to compare existing hallucination detection methods, evaluating not only how often models fail on BEAR-Bench but also how reliably those failures can be identified.

This paper investigates the feasibility of using residual learning to improve unsteady aerodynamic load prediction for aeroelastic applications. The machine learning technique selected for the study is the long short-term memory (LSTM) neural network, which is used for its suitability for sequential data with aerodynamic memory effects. The approach is investigated for the NLR 7301 airfoil benchmark using high-fidelity CFD lift data for prescribed pitch and plunge motions in the transonic flow regime in the presence of shock motion. An analytical unsteady aerodynamic model based on the Wagner function is used as a physics-based baseline, and the neural network is trained to learn the difference between the CFD lift coefficient and the Wagner prediction. The residual model is compared with a direct neural-network model trained to predict the CFD lift coefficient. The comparison includes feature and normalization studies, external benchmark cases, and leave-one-out and leave-family-out generalization tests across a range of sinusoidal and non-sinusoidal motions. The residual model performs best when its inputs align with the Wagner formulation variables, generally giving lower error and more consistent performance across training runs, though the direct model remains more accurate for some high-frequency cases. The residual model also generalizes better in the leave-one-out and leave-family-out tests, with a smaller increase in error than the direct model when entire motion families are withheld from training. Overall, the results indicate that residual learning shows promise as a modular approach for augmenting classical low-order aerodynamic theories, especially when the physics baseline removes a structured part of the aerodynamic response and leaves a lower-variance correction for the neural network to learn.

Reconstructing prompts that can elicit a desired answer or behaviour in an LLM is an open and important research topic. Optimisation methods which aim at minimising the perplexity of a given answer, however, consistently yield so-called pseudoprompts, unintelligible strings of tokens which can lack human interpretability. We argue that this is a consequence of the ill-posedness of the prompt optimisation task. By reframing the task as a Bayesian posterior inference over prompts, we propose an efficient algorithm to sample prompts which are both efficient (in terms of perplexity) and human readable. We compare our approach with state of the art alternatives showing on a real data set a marked improvement over a range of metrics.

Tabular prediction is a critical task across numerous applications. The recent success of large language models has sparked various approaches for adapting them to the tabular domain. A prevalent strategy involves training or fine-tuning specialized Tabular Foundation Models (TFMs) such as TabPFN. However, TFMs require substantial computational resources, and frequent model retraining is often impractical. In-context learning (ICL), specifically, few-shot prompting, offers a resource-efficient alternative to enhance performance. Yet, identifying the most relevant rows to serve as shots remains a challenge for tabular data. This paper introduces ARASH (Adaptive, query-specific Retrieval And Shot selection), a method that improves TFM efficiency by selecting optimal shots based on local neighborhood analysis within the training set. Our results demonstrate that ARASH reduces the prompt length and memory usage of TabPFN by 1261.5$\times$ and 2.56$\times$, respectively, while providing comparable accuracy.

Split federated learning (SFL) has emerged as a powerful paradigm for model training at the edge. However, SFL inherently involves discrete decision variables for model splitting and resource allocation, resulting in a challenging mixed-integer problem. Consequently, prior optimization schemes for SFL are either \textit{heuristic} or \textit{computationally inefficient}, which cannot handle large-scale user populations. To address this limitation, this work establishes an efficient optimization framework for SFL under resource-constrained networks. Our framework jointly optimizes model splitting and resource allocation to minimize training cost, which is defined as the weighted sum of latency and energy costs. We first study the model splitting problem and develop a polynomial-time algorithm that achieves the global optimum. Then, we extend the approach to the joint model splitting and resource allocation problem. In this case, we formulate it as a two-dimensional master problem and develop an efficient approximation method with a $(1+ε)$-approximation guarantee. Extensive experiments show that the proposed approach provides efficient solutions to strike the optimal energy--latency tradeoff.

Geospatial Foundation Models (GFMs) are emerging as a powerful paradigm for learning semantically rich and geographically consistent visual and physical representations. However, their reliance on Earth-observation (EO) data leaves information about human activity largely underrepresented. Human mobility data reveals the functional and relational structure between regions that is missing from EO data, but is often limited only to the city where it is observed, making it challenging to use for transferable urban representation learning. We introduce MoRAX, a lightweight framework for augmenting geospatial embeddings with functional structure derived from human mobility. MoRAX preserves the coverage and consistency of a GFM while providing information about the functional connectivity among urban regions, permitting zero-shot deployment in unseen cities with or without available mobility data. Across four target cities spanning two countries, the MoRAX teacher model, which observes mobility, consistently outperforms GFMs and strong urban representation baselines in eight socioeconomic and environmental prediction tasks. Meanwhile, the student model, which never takes mobility data as input, approaches the teacher in performance on most tasks. Transfer results across countries further demonstrate that modulation conditioned on mobility flows provides a general mechanism for grounding geospatial foundations in the human dimension of cities.

Large language models (LLMs) have demonstrated strong performance on structured reasoning tasks, but what they encode and whether it informs model behavior remain unclear. We investigate this question through geometric reasoning, using parametric CAD constraints as a controlled testbed for separating local pairwise relations from sketch-level constraint status. By probing the hidden states of six frozen decoder-only LLMs, we examine four properties: linear decodability, forced-choice generation, activation-level influence, and behavioral steerability. Pretraining substantially improves the decoding of local geometric relations, and this advantage persists after accounting for positional cues with shuffled-order controls. In contrast, sketch-level DOF status is already highly decodable from randomly initialized representations and improves only modestly with pretraining, indicating that much of its probe performance is available without learned weights. Further analyses show that decodable information is not always actionable. Generation often fails to express this information, and on the two intervention-tested backbones, activation-restoration effects at the patched entity position vanish while decodability persists across depth. Mean-difference steering also does not reliably control outputs. These results show that decodability, generation, activation-level influence, and steerability can diverge in the tested setting. The audit provides a controlled way to distinguish failures to encode geometric structure from failures to express or control encoded information.

Multi-armed bandit algorithms are evaluated by regret, yet comparable regret can coexist with different allocations across independent runs. We study the trade-off between worst-case regret $\mathcal{R}_{K,T}$ and instability $\mathcal S_{K,T}$, defined as the largest standard deviation of a terminal pull count, for $K$ arms and $T$ rounds. We prove the finite-time lower bound $\mathcal R_{K,T}\mathcal S_{K,T}\ge C T^{3/2}$, where $C$ is independent of $K$ and $T$, under a finite-time regret condition and without the regularity assumptions imposed in the prior asymptotic analysis. We also introduce Stabilized Lower-Envelope UCB (\textup{\textsc{SLE-UCB}}), a new tunable algorithm combining a running lower-envelope index with a decreasing pull-count stabilizer. \textup{\textsc{SLE-UCB}} satisfies $\mathcal R_{K,T}\mathcal S_{K,T}=O(T^{3/2}\log K)$, with an implicit constant independent of $K$ and $T$, matching the lower bound exactly in $T$ and within a logarithmic factor in $K$. To prove the instability bound, we develop a new offline top-prefix representation that removes path dependence from online decisions. Together with single-reward perturbations and the Efron--Stein inequality, this representation controls pull-count variance. Thus, regret and instability depend reciprocally on $K$, while their product has no polynomial dependence on $K$. These results resolve the open question raised in the literature concerning the sharp arm-dependent regret--instability frontier.

As large language models (LLMs) are granted increasing autonomy, it is essential to investigate methods that can induce unsafe behavior. We propose a novel white-box attack inspired by locate-then-edit approaches from the field of Knowledge Editing. Our choice is motivated by the observation that models edited with such schemes tend to assign unusually high prediction probabilities to the edit target, a property that is particularly advantageous when designing attacks. We modify the editing framework by incorporating as- sociative knowledge retrieved from the model, thereby extending constraint removal to an entire thematic category rather than being limited to prompts from a predefined dataset. Experiments with various archi- tectures demonstrate improved attack effectiveness over competing methods without dealing critical damage to general model performance.

Large language models are pushing data science toward increasingly autonomous and agentic workflows, with recent systems already supporting multi-step and long-running analyses. As these workflows become more autonomous, conventional interfaces no longer provide adequate support for two critical requirements: observability for understanding an agent's evolving reasoning and evidence, and steerability for redirecting low-value directions or deepening promising ones during execution. Existing interactive approaches improve process visibility and open intervention points, but they remain largely designed for discrete, turn-by-turn exchanges rather than the parallel branches and evolving decision structures of long-running agentic analysis. We study this need as interactive oversight in long-running agentic data analysis and present AdaLens, an interactive system for monitoring and steering ongoing runs. AdaLens combines a storyline-based representation that unifies analytical plans, execution progress, intermediate findings, and data-column involvement with steering interactions grounded in these analytical elements for directional guidance and execution control. We evaluate AdaLens through two case studies and a user study, examining how it supports analysts in monitoring and steering long-running agentic data analysis.

LLMs increasingly rely on external contexts, such as pre-defined system prompts or retrieved documents, to improve generation quality. However, processing these contexts alongside user queries creates an attack surface: adversarial inputs can induce models to disclose them. Prior probing studies suggest that leakage-related signals emerge in hidden states, yet the need to extract these states poses additional deployment challenges. In this paper, we explore whether this internal signal leaves a more accessible ``tell'' before decoding. We propose LeakGauge, which probes this response by appending a suffix that gauges leakage behavior and mapping its prefill token probabilities to an attack-risk score. While a direct gauge uses the initial tokens of confidential content, we find that a content-agnostic one that verbalizes leakage behavior yields more robust signals. Across 11 LLMs, including GLM-5.2 (753B) and Kimi-K3 (2.8T), LeakGauge reaches an AUROC range of 0.944--0.996 on unseen attacks. The signal remains stable when the content changes language or the attack shifts from verbatim to semantic disclosure. By activation-steering interventions, we further show that the risk score is sensitive to an internal leakage-related direction, relating the observable signal to the model's internal representation. In addition, LeakGauge enables an input detector with fewer than 0.5K extra parameters and added latency of 10.34 ms. Code: \href{https://github.com/yeasen-z/LeakGauge}.

Public institutions face a persistent challenge in selecting LLMs suited to their specific context. Existing benchmarks, however, are of limited use as they primarily reflect English-language and US-centric settings, and often only evaluate task performance. In this paper, we present first results of MÖVE, a holistic evaluation framework for the German public sector, examining three rarely considered governance dimensions: energy consumption, provider transparency, and knowledge of German-party positions. Our results reveal significant trade-offs, with no single model excelling across all dimensions: estimated energy consumption varies more than 60-fold and is not explained by model size alone, information disclosure varies systematically across providers, and European models do not exhibit stronger knowledge of German party positions. Model selection for public institutions thus cannot rely on performance rankings alone. Instead, evaluations should also reflect the governance requirements of the deployment context.

The traditional Software Development Lifecycle (SDLC) assumes that system behavior is determined primarily by source code, allowing correctness to be specified, implemented, and verified through code-centric practices. Data-intensive and AI-enabled systems challenge this assumption because their behavior emerges from the interaction of code, data, and learned models, while performance may degrade as real-world conditions drift from training data. This paper examines how integrating data engineering and software engineering practices, operationalized through DataOps, MLOps, and LLMOps, reshapes the SDLC for these systems. We make four contributions. First, we synthesize literature across software engineering, data management, machine learning systems, and human-centered computing into a phase-structured account of lifecycle transformation spanning requirements, architecture, development, testing, deployment, monitoring, governance, and organization. Second, we provide a lightweight formalization in which system behavior is defined over code, data, and model configurations; requirements become evaluation-led specifications with probabilistic acceptance regions; and promotion is controlled through statistically grounded validation gates. Third, we develop an adaptive five-layer lifecycle framework comprising artifact, contract, gate, control, and governance layers, positioning maintenance as a closed-loop control problem under configuration drift. Fourth, we propose a conceptual research model linking data engineering integration to measurable lifecycle outcomes and critically assess the evidence base. While the direction of transformation is increasingly established, its magnitude remains insufficiently quantified. We conclude with a research agenda for an empirically grounded, adaptive SDLC for data- and AI-centric systems.

Powered two-wheeler riders face critical safety challenges in low- and middle-income countries, yet limited studies exist on how cognitive stressors such as Time Pressure influence collision risk. To address this gap, we introduce a large-scale dataset of over 129,000 labeled multivariate time-series sequences from 153 simulator rides by 51 participants under No, Low, and High TP, capturing 64 features across vehicle dynamics, control inputs, proximity, and behavioral violations. Building on this dataset, we propose MotoSafety, a novel edge-AI architecture grounded in the Learned Temporal Importance principle. MotoSafety achieves 94.97% accuracy and 99.33% ROC AUC, outperforming ten baselines, including TimesNet and LLM4TS, and achieves 0.039 MSE and 0.094 MAE for forecasting (4.4x lower error than Time-LLM and iTransformer). With only 1.15M parameters and 0.135 ms latency, it is suitable for edge deployment on low-cost CPU hardware. Using ground truth TP as an inductive bias improves accuracy from 94.09% to 94.97%, while predicted TP achieves 94.82%. Using only 21 IMU+GPS features, it achieves 93.91% accuracy, indicating practical deployment. Beyond PTW safety, the architecture shows better transferability to human activity (97.66%) and clinical (99.65%) domains. This lightweight framework advances PTW collision risk assessment, supporting the Safe System Approach for Intelligent Transportation Systems.

Accurate building height information at the individual footprint scale is essential for material stock accounting and post-disaster damage assessments yet remains difficult to obtain at city scale in the Global South where airborne LiDAR coverage is rare and commercial very high-resolution imagery is cost-prohibitive or unavailable. While recent works have demonstrated building height estimation using freely available Sentinel imagery, the resolution ceiling of resulting products is still coarse for material stock analysis. This study incorporates products derived from data freely accessible under scientific research licenses, TerraSAR-X StripMap and PlanetScope, alongside Sentinel-1 to predict building heights in a large city in Brazil. To account for the spatial autocorrelation in the training set, features from all sources are integrated in a geographically weighted random forest model, returning an RMSE of 5.34 m and R2 of 0.756 against a LiDAR reference dataset. Local feature importance showed predictor dominance to vary consistently across intra-urban contexts, with footprint geometry dominating for low-rise buildings, shadow-derived height for taller and more isolated structures, and spectral reflectance for the tallest buildings in the set. Sentinel-1 backscatter and InSAR occupy complementary spatial niches, with no single sensor uniformly preferable across the set. Results provide optioneering guidance and insight over satellite-derived products predictive relevance in distinct contexts, which global machine learning or neural network models cannot offer.

The evaluation of large language models (LLMs) relies heavily on human-designed assessments, implicitly assuming that AI and humans employ similar underlying cognitive constructs. Challenging this assumption, we investigate whether the latent factors governing LLM performance carry the same substantive, human-interpretable meaning as the cognitive constructs governing human learners. Using responses from humans and six LLMs across quantitative reasoning and chemistry assessments, we conducted Exploratory Factor Analysis (EFA) separately for both groups. Subject-Matter Experts (SMEs) then blindly evaluated the resulting factor graphs to ascribe pedagogical meaning to the emerged constructs. SMEs successfully interpreted most of the human-derived factors. Conversely, they could not ascribe meaning to any LLM-derived factors in quantitative reasoning and interpreted only half of the LLM factors in chemistry. By combining data-driven EFA with blind expert interpretation, this framework shows that LLMs frequently operate on statistically opaque mechanisms distinct from human reasoning.

Humans naturally form and express beliefs in daily communication, e.g., "I think the answer is 3" or "I suppose that's right." Such beliefs inevitably intertwine with fact and knowledge, making the ability to handle them in tandem desirable for large language models (LLMs), as they are increasingly deployed in user-facing settings. Prior work showed that even capable LLMs exhibit a systemic weakness in acknowledging user beliefs grounded in incorrect information. We extend this evaluation to 10 LLMs across 18 epistemic expressions and find that the size and direction of the weakness depend on the verb used to express the belief, with the accuracy gap between factual and false information ranging from +50% on "I vaguely remember" to -14% on "I seriously doubt". We further show that the phenomenon stems from task confusion: models default to fact-checking the underlying claim, overriding the user's stated belief; chains of thought that explicitly fact-check show lower accuracy on false information than those that do not; and a single instruction can reverse the failure across verb families. Mechanistically, models attend more to false beliefs they fail to confirm, but suppressing this attention at decoding time recovers accuracy only partially and only in some models, calling for future work on intervention methods. Our findings clarify prior results and show how fact-checking, a generally desirable behavior, can interfere with belief tracking in LLMs. Our code is available at https://github.com/ngqm/belief-fact-phrasing.

Practical LLM unlearning is usually evaluated through two objectives: suppress target-specific knowledge and preserve non-target utility. In generative QA, this leaves a third behavior underspecified: when a target-adjacent prompt admits a broader answer without target-specific leakage, the model should answer at that level rather than leak, evade, or refuse. We study this specification problem in a controlled LoRA-GRPO RWKU setting, comparing four reward designs that span lexical suppression, anti-refusal shaping, rubric-based broad answering, and an explicit refusal contrast, with and without SFT warm-up. The experiments show that optimization success is not equivalent to behavioral unlearning: RWKU forget scores, held-out completion audits, terminal training-rollout audits, and training dynamics can point to different conclusions. We trace these disagreements to reward-hacking endpoints, policy-support limits in GRPO, benchmark probes that miss endpoint changes, and rewards that can select broad-topic answering with low semantic leakage during optimization.

Let $\varepsilon_1,\ldots,\varepsilon_n$ be independent Rademacher signs and let $a=(a_1,\ldots,a_n)\in\R^n$ satisfy the normalization below. For the normalized Rademacher sum, we determine how its higher moments depend on the fourth-order mass. Combining a sharp fixed-q moment envelope with a separate argument below the convexity threshold gives the Gaussian stability inequality for the full range $p\geq4$ of this linear-in-q bound. The same fourth-order framework determines the sharp finite dimensional $L_p/L_4$ Khintchine constant for $p\geq5$, with the flat coefficient vector as the extremizer. These results settle the conjectures of Jakimiuk and of Barański, Murawski, Nayar, and Oleszkiewicz stated below. We also prove Jakimiuk's conjectured quadratic stability estimate at $p=3$. The resulting bounds retain information about sparsity and effective dimension, with applications to Rademacher random projections and randomly signed errors; those applications are not developed further here. Their Laplace-transform form also gives coefficient-sensitive tail bounds. The proofs are discovered with substantial assistance from ChatGPT 5.6 Sol.

Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually demand from AI systems. We introduce \textbf{StartupBench}, an E2E agent benchmark grounded in market-validated AI startup products. Rather than defining tasks from pre-defined assumptions about useful agent capabilities, we systematically study AI products with demonstrated adoption, together with their product workflows and users, to identify real-world tasks for which AI has established practical demand across diverse professional domains. We translate these workflows into complete deliverable-oriented tasks and evaluate them with fine-grained rubrics capturing their complex requirements. Across representative models evaluated under a unified agent harness, even the strongest model successfully completes only approximately 30\% of StartupBench, despite making substantial partial progress on many tasks. Further analysis identifies aspects like complex instruction following and domain-specific expertise as major sources of failure. Our results reveal that many market-validated workflows remain beyond the reliable capabilities of current general-purpose agents, establishing StartupBench as an empirical measure of progress toward E2E completions of real-world user tasks.

Ground-to-Air (G2A) drone detection in medium- and long-wave infrared (MWIR/LWIR) imagery is challenging due to reduced texture information, sensor noise, weak thermal contrast, and the scarcity of annotated data. This work investigates a synthetic-first training strategy that combines synthetic scene generation with fine-tuning on real data. We show that synthetic data provides an effective basis for learning initial object representations, while real in-domain thermal imagery is still essential for reliable deployment. Even small amounts of real IR data substantially reduce domain gaps. Our experiments indicate that dataset alignment has a stronger impact on performance than model scale. Finally, our analysis of the dataset suggests that semantic alignment in feature space is the strongest predictor of model performance, while radiometric properties such as entropy and dynamic range also contribute to detection robustness. This work provides a foundation for combining synthetic and real IR data for effective G2A drone detection.

Deep learning-based DDoS detectors for 5G-enabled cyber-physical systems face scarce labeled attack data and unrealistic synthetic substitutes, which limit robustness against adaptive adversaries. Detectors trained on hand-crafted attacks with fixed scaling multipliers degrade catastrophically (F1-score drops of about 47 percent to 100 percent, depending on scenario) when confronted with realistic, distribution-preserving samples. We propose Diff-DDoS, a three-phase framework for realistic attack synthesis and robust detection using tabular diffusion models. Phase 1 trains a baseline CNN cell-level detector on spatiotemporal grids from call detail records (CDRs). Phase 2 trains a tabular denoising diffusion probabilistic model (TabDDPM) on normal CDR aggregates to generate realistic attacks and expose detector vulnerabilities. Phase 3 introduces adversarial diffusion training (ADT), using inverse classifier guidance to generate hard yet distribution-preserving samples until the detector converges. On a Milano CDR dataset across SMS-flooding, silent-call, Internet-signaling, and blended scenarios, ResNet50 with ADT recovers F1-scores of 79.62 percent (silent-call), 100 percent (Internet), and 92.79 percent (blended). After validation-based threshold calibration, ADT reaches 100 percent SMS F1 versus 47.3 percent for CTGAN, and matches the strongest gradient-based adversarial-training baseline on silent-call. These results support tabular diffusion models for stress-testing and hardening intrusion detectors in data-scarce 5G cyber-physical deployments.

Text-to-SQL systems are commonly evaluated using ground-truth SQL queries or reference execution results, but such supervision is unavailable at inference time in real-world deployments. This creates a critical verification problem: given only a user question, database context, and generated SQL, can a system estimate whether the generated query is likely to correctly answer the question? Recent approaches use LLMs as judge or specialized agents to inspect generated SQL, but their decisions can be difficult to trace. Outcome Reward Models (ORMs) address this by learning from execution-labeled candidate SQLs and assigning correctness scores to unseen queries, yet they still provide limited visibility into the signals behind each verification. To address this limitation, we propose TraceSQL, a lightweight and traceable verification model built on explicit diagnostic features. TraceSQL combines 67 features capturing question ambiguity, question requirements, question-schema-SQL consistency, SQL structure, and intent alignment. These signals remain available for examining which factors influence each prediction and for tracing decisions back to diagnostic evidence. On BIRD development databases, TraceSQL achieves 66.47% F1 and 64.48% ROC-AUC, compared with 61.87% F1 and 58.26% ROC-AUC for the GradeSQL-7B ORM baseline on the same generated-SQL evaluation. Feature attribution further shows that the model relies on both semantic grounding and deterministic SQL-structure signals. These results show that SQL verification can be performed with a lightweight learned model while retaining feature-level evidence for inspecting and diagnosing its predictions.

ML systems increasingly condition decisions on downstream model identity, but this is useful only if model-specific differences form reusable structure rather than input-local interactions. We test this in retrieval-augmented generation (RAG), where evidence utility can be measured under controlled interventions. Holding query, evidence, task, scoring, and intervention fixed, nine readers disagree on effect sign in 33\% of jointly affected cells; reader$\times$query interaction explains 29.8\% of utility variance versus an 8.4\% permutation null; and self-selected evidence improves F1 by $+0.031$ ($t=3.39$). We then ask the sharper question: \emph{which components of this heterogeneity are stable reader properties across queries?} Separating three measurable objects---evidence \emph{activity}, \emph{ordinal preference}, and \emph{conditional signed direction}---we find ordinal reader geometry stable across four independent settings (split-half $ρ=0.60$--$0.83$): leave-one-out interventions, PRISM preferences, RAMDocs, and RAGuard. Signed geometry is task-bounded: weak in open-ended QA (0.14, 0.35), especially for misleading and irrelevant evidence, but strong in binary fact-checking (0.75) with no significant ordinal gap, though still below its sparsity-matched ceiling. Sparsity, decoding noise, and metric artifacts do not explain the main ordinal--signed gap. Finally, stable ordinal similarity fails to predict cross-reader intervention transfer (oracle-distance $ρ=-0.27$; regret reliability $-0.28$). Reader-specific utility exists, but preference is not intervention: stable ranking similarity does not license transfer of help/harm decisions.

We demonstrate that RL finetuning an LLM using debate, a two-player adversarial game between a generator and a critic adjudicated by a weaker LLM judge, reduces reward hacking compared to a reinforcement learning from AI feedback (RLAIF) baseline. Reward hacking is a central obstacle in RLAIF: as training progresses, the policy learns to exploit systematic errors in its AI judge, degrading task performance, a problem that worsens precisely when the judge is weaker than the policy, the setting most relevant to overseeing increasingly capable AI systems. We study mathematics tasks, where final-answer correctness is verifiable, allowing us to measure reward hacking dynamics. We train a Gemini~2.5 Flash-class policy with a frozen, weaker Gemini~2.5 Flash Lite judge, comparing a single-player RLAIF baseline against debate. While the baseline quickly hacks the judge, debate maintains judge performance throughout training, leading to a higher peak validation accuracy (45\% performance gap recovered) that persists through many RL steps. Additional experiments show that: 1) further weakening the judge leads to faster hacking, but this can be compensated by adding an additional debate round; 2) debate incentives override prompted misalignment; 3) RL using an LLM judge has a smaller train/validation reward gap than RL from verifiable rewards; 4) learning to critique to convince the judge using ground truth labels is possible but slow. Taken together, our results are a positive update on the feasibility of debate, while highlighting that balancing multi-agent training is critical: without player constraints, adversarial training risks defaulting to critic judge-hacking. We show that critique word limits (effective up to 150 words) successfully balance the game and avoid judge hacking, though this introduces a trade-off by restricting critic expressive clarity.

Anomaly detectors are hardest to deploy exactly where training data is scarcest: a newly commissioned production line has a handful of verified "golden" samples and no machine-learning engineer on the factory floor. We present a training-free human-in-the-loop framework in which a domain expert corrects a PatchCore detector by direct memory bank editing: no retraining, no gradients, no original training data. A false-positive correction inserts the reviewed image's normal patches through a self-calibrating novelty gate admitting only those beyond the median pool-normal nearest-neighbour distance. From a bank built on only ten golden samples, operator corrections close a median 66% of the gap to an uncorrected fully trained bank (mean 80%, raised by three categories that overshoot parity), significantly improving 12 of 15 MVTec AD categories and harming none: ten samples plus corrections outperform hundreds of samples without them. On already-trained banks the headroom is smaller and concentrated where the bank undersamples normal appearance (gated: toothbrush +0.10, metal nut +0.09, zipper +0.05, screw +0.05), and no category except grid is significantly harmed. Evaluation uses a held-out protocol (20 splits per category, Holm-corrected Wilcoxon), because corrected images entering the bank inflate naive evaluation toward AUROC 1.0 by memorisation. Passive and active querying are statistically indistinguishable; a matched-label-budget control attributes gains to deployment-time label production at 43% of exhaustive-review cost; a defect-memory extension fails decisively. Feedback is simulated from ground truth; live expert trials, where mislabelling is costliest on small banks, remain future work.

Intelligent channel state information (CSI) feedback is essential for realizing the high capacity and spectral efficiency goals of future 6G systems, yet existing deep learning solutions face a trade-off between model generalization and scenario-specific performance. Large neural networks generalize well but incur high computational and tuning costs, while small models excel in particular environments but require repetitive costly end-to-end training for each base station (BS). To address these challenges, we introduce a model repository-based deployment framework in which a centralized AI data center maintains a catalog of scene-specific CSI models. The repository is enhanced with a Learnware-based framework, where each model is associated with a specification including semantic part (network architecture parameters) and statistical part (codeboo-fingerprint embeddings of training-data distributions). A BS submits only its local statistical specifications to retrieve the most relevant pre-trained model, enhancing data privacy by avoiding raw CSI transmission and drastically reducing retrieval latency and communication overhead. We further develop a data-driven search strategy that matches codebook fingerprints to model performance, achieving over 90% selection accuracy. In simulations, our scheme yields 18.8% and 57.7% performance improvements over the General Model in LOS and NLOS scenarios, respectively while reducing local fine-tuning by up to 1000 samples and 100 epochs. This Learnware-based approach minimizes redundant training, maximizes model reuse, and supports rapid,privacy-enhancing deployment of CSI feedback models.

Memory is a key capability of LLM agents. Persistent memory extends this across sessions---enabling recall, revision, and personalization. Yet its multi-stage pipeline (ingestion, retrieval, filtering, generation) makes failures difficult to localize: end-to-end evaluation reveals that an error occurred, but not which stage caused it. Existing evaluations often report aggregate performance without paired statistical comparisons, slice-level non-regression checks, or stage-level diagnostic traces. We propose D$^2$ACCI (Diagnostic-Driven Artifact-based Closed-loop Controlled Iteration), a dual-loop protocol whose outer diagnostic gate promotes, feature-flags, or rejects memory interventions based on paired evidence, protected-slice monitoring, and trace-level localizability. We further introduce DCR, a graded observability metric that measures whether failures remain localizable, and D$^2$ACCI-Eval, a reusable artifact for gate replay. We instantiate the protocol in MemStack and evaluate on three public benchmarks, achieving 93.59% on LoCoMo, 90.93% on LongMemEval, and 57.20% on PersonaMem-V2. Five paired ablations show that supplement extraction, session-memory retrieval, and Forget Guard yield statistically significant gains (+1.9 to +3.7pp, all p $\le$ .003). In contrast, BM25/RRF is retained as a monitored feature flag---a distinction invisible to aggregate-only evaluation. A diagnostic audit shows enriched traces substantially improve root-cause agreement over result-only relabeling. Diagnostic artifacts reach 98--100% DCR@3 versus 0% for results-only logs. These results establish that robust memory-system iteration demands traceable, statistically grounded, and regression-aware evidence---exactly the gap D$^2$ACCI fills.

Short-duration heavy-rainfall warning determines whether 1 h rainfall will exceed a threshold within a target-station neighborhood over the next few hours. Multitemporal infrared and water-vapor observations from the Fengyun-4A Advanced Geostationary Radiation Imager (FY-4A AGRI) capture cloud-top cooling, moisture evolution, and cloud expansion before substantial surface rainfall develops. However, most deep-learning nowcasting methods convert these signals into local warnings by post-processing gridded precipitation predictions, preventing station-neighborhood event targets from directly supervising the satellite-to-station learning pathway. We propose MAGPIE-Net, which embeds a geographically adaptive, differentiable grid-to-station mapping in a pathway combining convection-initiation features, multiscale encoding, and auxiliary gridded precipitation diagnosis. Station-neighborhood event losses thereby constrain the satellite representation and its mapping to irregular station locations for 0-3 h event prediction. In independent 2023 warm-season tests over central and eastern China, critical success index (CSI) values under the primary 40 km/20 mm h-1 definition were 0.371, 0.304, and 0.238 at 0-1, 1-2, and 2-3 h. Across episodes, MAGPIE-Net achieved a detection rate of 65.1% and a mean lead time of 64.6 min, compared with 23.6% and 18.3 min for the best gridded-output baseline, and remained superior for smaller neighborhoods and the 50 mm h-1 threshold. During the critical early-warning stage, when antecedent 1 h rainfall within 40 km remained below 1 mm, MAGPIE-Net detected 51.9% of episodes with a mean lead time of 38.5 min. These results show that event-oriented satellite-to-station modeling converts multitemporal geostationary cloud and moisture observations into local heavy-rainfall warnings more effectively than gridded-precipitation modeling.

Decentralised partially observable Markov decision processes (DecPOMDPs) provide a general framework for modelling multi-agent decision making under uncertainty. However, DecPOMDPs are known to suffer from exponential complexity in the number of agents. One way to combat this intractability in agent numbers is to look at partitions of agents that exhibit a form of symmetry among agents, allowing for a compact encoding by counting. However, a challenge arises as the policy space explodes, even though the model complexity and evaluation cost reduce to a polynomial dependence. In this paper, we redirect our focus from counting agents to counting policies, which actually enables tractability in agent numbers for so called policy-counted DecPOMDPs. Further, we present policy-counted dynamic programming using the compact representation to solve policy-counted DecPOMDPs efficiently.

Take three frontier mixture-of-experts models (Alibaba, OpenAI, NVIDIA; 3.6-4.0B active parameters each) and fine-tune them to reason in a low-resource language. On accuracy benchmarks almost nothing happens, and the benchmark itself is noise at this scale: changing only the random seed moves the score by 7.7 points, more than every data and recipe effect we measured. That null is our first result. The real changes live where accuracy cannot see. Base models never think in Greek: 0 of 1,000 reasoning traces, even when the question is Greek, so the model answers correctly while reasoning in a form its user cannot read, audit, or correct. After supervised fine-tuning (SFT), every released checkpoint reasons in the language of the question on ~98% of items, one family at 3x fewer tokens, with judged grammaticality improving on all four models and general ability within a few points of each base: nothing was forgotten, and fluency was gained. We propose six behavioural dimensions that make such changes measurable, each gated to reject any metric that correlates with output length, and we report how our own instruments lied: six failures, each caught by a control. What SFT cannot do is fix its own defects: a quarter of answers skip the requested format, answers leak into the reasoning channel, and an explicit "think in English" is obeyed under half the time. Reinforcement learning with verifiable rewards, pre-registered before training, fixes the first two outright (fallback 24% to 2.5%, leak 3.5% to 0.0%, both against a flat random-reward control) and moves the third (+9.1pp), while the Greek reasoning habit survives an accuracy-only gradient untouched. We release five checkpoints. The instruments, the controls and the pre-registration travel to any low-resource language; Greek is the case that let us measure them.

OWL 2 DL ontologies, grounded in the description logic $\mathcal{SROIQ}$, express large knowledge bases in biomedicine and the Semantic Web. Neuro-symbolic (NeSy) learners over description logics either embed the ontology in a continuous space, abandoning classical entailment, or restrict to the Horn fragment $\mathcal{EL}^{++}$, which has a single canonical model. We present Baobab, which compiles a $\mathcal{SROIQ}$ ontology with a finite ABox into a Sentential Decision Diagram (SDD): it saturates a propositional core under a consequence-based calculus and instantiates the remaining $\mathcal{SROIQ}$ features (nominals, number restrictions, and the role axioms) over the active domain. The SDD's evidence-conditioned weighted model count then trains a perception network to recognize real images under partial ABox supervision: on an ontology that exercises every distinctive $\mathcal{SROIQ}$ feature, a CNN learns to read MNIST digits coupled by a successor relation and recovers latent ontology concepts that an independent perception leaves at chance. When the supervision admits several ontology-consistent completions, an independent perception collapses onto one, a reasoning shortcut: we show that a mixture indexed by the query's justifications can represent the calibrated posterior no independent perception can, and that seeding it from the circuit's enumerated completions attains the Bayes-optimal posterior on a real-image MNIST task where single-WMC and learned mixtures (the BEARS-ensemble hypothesis class) do not: to our knowledge the first to characterize and mitigate reasoning shortcuts in a non-Horn description logic. Soundness of the compiler and the representation result are machine-checked in Lean 4. Code is available at https://github.com/bio-ontology-research-group/baobab.

Ethereum's consensus safety relies on independent consensus client implementations agreeing on every state transition. When they diverge due to implementation errors, the network can fork, finality can stall, and severe attacks are possible. To prevent such consensus divergences, Ethereum provides a Python reference implementation (consensus-spec), which acts as a specification, and a hand-crafted official test suite (spectests). However, as an executable implementation, Ethereum's specification defines validity implicitly through runtime behavior. As a result, it lacks a systematic way to ensure that all validity conditions are thoroughly evaluated. We present SpecTrum, a framework that addresses this problem in three stages. First, we introduce Consensus-SpecTec, a mechanized specification of the Ethereum consensus algorithm, which makes validity conditions explicit as if-premises. Second, we define premise coverage, a metric that measures which if-premises are evaluated to true and false across spectests. Third, we develop a specification-based test generator that extracts constraints on premises not evaluated to false by spectests and generates inputs to evaluate them. Applying SpecTrum to five major Ethereum consensus clients, we identify 27 cross-client divergence cases, 22 of which cannot be found without the premises inserted in our mechanization. All 27 cases reproduce across fork versions, and extending the mechanized specification to a new fork takes modest effort proportional to the specification difference.

Diversity is a fundamental criterion for evaluating generative artificial intelligence (AI) systems, yet its measurement remains inherently ambiguous. Existing approaches typically represent generated samples in an embedding space, compute pairwise distances or similarities, and aggregate them into a single scalar score. Such scalar summaries are convenient, but they often encode different inductive biases and may yield contradictory rankings of the same sample sets. In this paper, we argue that diversity evaluation for AI-generated content is intrinsically under-specified when reduced to a single number. We first review representative diversity metrics, and then diagnose their limitations from two complementary perspectives: an axiomatic analysis showing that no representative scalar metric satisfies all desirable properties simultaneously, and an empirical analysis showing that high-dimensional representation spaces can induce concentrated, modality-dependent distance distributions. To address these issues, we propose diversity profiles: curve-valued, condition-aware summaries that evaluate a parameterized diversity family across a range of thresholds, scales, exponents, or orders under a specified representation and distance or kernel function. Diversity profiles reveal whether a comparison is robust across resolutions or instead depends on an arbitrary parameter choice. We instantiate profiles for several representative metric families and demonstrate their practical use in generative AI evaluation. Overall, diversity profiles provide a more transparent and resolution-aware framework for comparing the diversity of AI-generated content.

Vision-Language models (VLMs) achieve outstanding performance largely due to the amount of training data available on the internet. At the same time, data holders (e.g., artists) urgently need to determine whether their data has been used for model training without authorization, which concerns both intellectual property rights and personal privacy. Data auditing, particularly through membership inference (MI), has attracted attention as a direct tool. This work proposes MemCatalyst, a set of data poisoning tools, aiming to amplify the data auditing performance on VLMs. MemCatalyst employs two strategies: Poisoning Text (PT) and Poisoning Image (PI). MemCatalyst forces VLMs to over-learn specific inconsistencies between image features and textual semantics during training, thereby increasing their susceptibility to membership information auditing. Crucially, the transferability of poisoned samples across different VLM architectures is demonstrated to be effective in the black-box setting. Extensive evaluations using five state-of-the-art data audits on two prominent VLMs demonstrate that MemCatalyst markedly enhances MI AUC scores with a minimal budget of poisoned samples, while maintaining a negligible impact on model performance.

Context: Software systems that depend on commercial large language model APIs must migrate to successor versions when vendors deprecate older models. Migration decisions typically rely on aggregate benchmark scores, which compress heterogeneous item-level behaviour into a single net figure. Objective: We measure what that compression conceals. Method: On three pairwise upgrades in the GPT-5.4 to GPT-5.6 Sol product sequence, we query 900 public benchmark items (graduate-level knowledge, olympiad mathematics, instruction following) 50 times per item per model, classify each item as reliably improved, reliably regressed, practically equivalent, or inconclusive under false-discovery-rate control and a practical-significance threshold, and calibrate the results against a label-permutation null. Results: Across all nine migration-benchmark cells, reliable improvements and reliable regressions coexist. Edges with aggregate gains of up to 7.3 percentage points contain up to 8.3% reliably regressed items; edges with aggregate losses contain up to 10.7% reliably improved items. On the instruction-following benchmark, the gap between strict and loose scoring widens by 3.9 percentage points on the latest migration: a 3.9-point regression under strict scoring shrinks to 0.04 points under loose scoring. Conclusion: Migration decisions based on aggregate scores alone miss substantial bidirectional item-level change. The complete response-level archive and per-item scoring outputs are released.

Long-horizon agents increasingly operate across many steps, tools, and observa- tions. In this setting, the relevant oversight question is not only whether each action is locally valid, but whether the evolving trajectory still corresponds to the task the user authorized. Drift can accumulate quietly: an agent may call the right tool with plausible arguments at every step, while its prefix moves toward a broader role, an adjacent objective, or evidence the user never supplied. Existing monitors mostly check local compliance, deliver final-trace verdicts, or score generic risk; they do not directly estimate this prefix-level relation. We introduce ontological trust, a task-conditioned property of trajectory prefixes, and instantiate it as RGE, an online monitor that decomposes trust along Role, Goal, and Evidence. RGE uses LLMs only to derive structured task and step representations; trust-state updates, projec- tions, and intervention decisions are deterministic, so the output is a replayable and auditable trust trajectory rather than a single end-to-end judge verdict. We construct a cross-domain trajectory corpus from OSWorld, FinanceBench, and EICU-AC, covering benign executions, prefix-paired drift, and pseudo-consistency failures. On this corpus, RGE outperforms adapted rule-, judge-, and shield-style baselines on prefix-paired drift detection. With the two larger estimator models, it exceeds 93% Drift F1 on every benchmark while keeping benign coverage at or above 95.8%. Pseudo-consistency is harder: detection depends on whether task completion is externally visible, a structural limit we characterize empirically.

Credit decisioning is a high-stakes task in which model outputs must be accurate and explainable to support compliant decisions. Although modern credit risk models such as eXtreme Gradient Boosting (XGBoost) and Graph Neural Networks (GNNs) improve predictive performance, their explanations are often too technical for stakeholders creating communication gaps that can shape approvals, denials, and fairness judgments. We examine whether Large Language Models (LLMs) can serve as explanation layers that translate post-hoc explanation artefacts into stakeholder-appropriate risk narratives. Using Freddie Mac single-family loan-level data, we develop three pipelines: standard tabular (XGBoost + SHAP), and two with alternative data, a pure network-based (GNN + GNNExplainer), and a bimodal one (combining tabular and network data). We generate narratives with three LLM configurations: a small fine-tuned LLM (Gemma 3 4B), a large fine-tuned LLM (DeepSeek R1 70B), and a zero-shot commercial LLM (Gemini 2.5). Explanation quality is evaluated through automated checks across all pipelines and a human study of bimodal explanations comparing credit risk professionals and non-professionals on eight decision-relevant dimensions. We have three main findings. First, the pipeline accounts for higher variance in evidence-grounding scores than the language model, meaning that the binding constraint on explanation quality is the evidence representation, not the model used. Second, the explanation narratives reliably name the influential factors but are less reliable when stating the direction of influence, which may be consequential for adverse-action communication. Finally, professionals apply stricter evidentiary standards than non-professionals. We discuss implications for the governance of risk models, including deployment considerations and the value of domain-aligned LLMs in regulated credit settings.

Agent evaluations and trace-based learning often compare outputs across transformed views through a post-response correspondence treated as neutral preprocessing. We show that this correspondence is a measurement intervention: omitting it can manufacture sensitivity, an over-aggressive map can manufacture invariance, and multiple optimal correspondences can leave mechanism labels and signed learning credit unidentified. We develop a validity theory and audit with three components: two-sided validation of nuisance removal and response preservation, all-optima identification of downstream conclusions, and uncertainty propagation after validity is established. We characterize the linear feasibility boundary for response-preserving nuisance removal, compute sharp ranges over exact-optimum correspondence sets, and give a distribution-free certificate that retains a credit coordinate only when all exact optima agree on its nonzero sign. Across public code and SQL pipelines, two deterministic optimal tracebacks disagree on temporal localization for 55.9% of 1,586 nonzero trajectory pairs; two frozen 800-rollout tool-use audits, including a task-and-seed-disjoint replication, expose exact-optimum reversals of intended turn-level credit, although a clean public quick-start subset shows none. A pre-registered transport gate failed on natural responses; frozen corrected and held-out controls then show that a map calibrated only on benign examples erases every retained harmful response, while two-sided validation selects response-preserving alternatives. Cross-view correspondence must therefore be declared, validated, and propagated into uncertainty before agent evaluation or credit assignment supports a point conclusion.

Prediction cascades significantly reduce energy consumption of Artificial Intelligence (AI) models while maintaining high predictive performance. The idea is that easy inputs are routed through a lightweight small model, and difficult uncertain cases are deferred to a larger model. While this design can improve computational efficiency on clean data, its effectiveness depends on the reliability of confidence-based routing. Input degradations, such as static corruptions and sequential perturbations, can shift model confidence and routing decisions. In this paper, we study confidence-based cascade frameworks for image classification and investigate how such degradations affect their confidence-based deferral behavior. We select a model cascade at the pareto-optimum of accuracy, routing quality, and energy consumption that achieves competitive predictive performance with an up to 10-fold decrease in CO$_2$ emissions. We study the behavior of that model cascade under input corruptions and analyze how the cascade's routing decisions change when the input distribution shifts. Our analysis identifies three failure modes. Static corruptions either (1) break the routing signal while the large model remains useful, or (2) degrade both models so deferral no longer recovers accuracy. Sequential perturbations reveal a third mode: predictions stabilize but deferral suppresses, yielding stable but unreliable predictions. These findings demonstrate that energy efficient model cascades require evaluation beyond clean accuracy, with explicit attention to routing reliability under distribution shift.

Mobile robots that operate in side by side with humans and critical facilities must reach their goals at low cost, despite often unknown true traversal costs of the map apriori and imperfect actuation. Planners that solve the underlying stochastic shortest path problem exactly, such as value iteration, require computation that grows with the diameter of the map, whereas Dijkstra's algorithm is fast but is usually considered inexact once transitions are stochastic. This study shows that Dijkstra's algorithm can remain an exact planning engine under a condition that is much weaker than the causality condition often invoked in the literature, namely nonnegativity of a reduced cost defined on the determinized map. Building on this characterization, an online learner DORA (Dijkstra Oracle Reduced-cost Algorithm) is proposed for robot navigation that calls a shortest path oracle a fixed number of times per episode, never estimates a transition kernel, and adds a logarithmic survival weight when the probability of contact with a dynamic obstacle must stay within a budget. In the numerical experiments involving three other benchmarks that cover grid world navigation, directional drilling, and drone surveillance, the learner matches optimistic value iteration that is given the true transition kernel while performing 4.5 to 19.3 times less planner work, reduces contacts during learning by a factor of seventeen relative to determinize and replan, and keeps the contact rate within budgets that span two orders of magnitude. These results indicate that shortest path search supports safe and efficient online navigation and path planning tasks.

Existing LLM-based approaches to Architecture Decision Record (ADR) generation share a critical and largely unexamined assumption: that input is already reasonably structured. In practice, architectural decisions emerge from informal, noisy meetings where choices are implicit, fragmented, and entangled with off-topic dialogue, precisely the conditions under which single-pass prompting degrades. This paper presents GADR, a multi-agent, self-correcting workflow that extracts architectural decisions from raw meeting transcriptions and generates Nygard-formatted ADR drafts. A feasibility study comprising five real project meeting transcripts, expert review by four senior architects, and evaluation by fifteen students provides initial evidence that the agentic workflow captures most expert-identified decisions and produces drafts participants found clear and useful, outperforming zero-shot and few-shot baselines in stability and structural adherence. The study also addresses the underexplored trade-off of RAG-based enrichment improving ADR depth while simultaneously risking transcript-unfaithful content, raising open questions about traceability in automated architectural documentation that we believe is worth the community's attention.

Despite their widespread use, Large Language Models (LLMs) remain limited by a fundamental problem: the generation of plausible but false content, known as hallucinations. Most existing detection methods operate at the answer or sentence level, yet per-token detection is essential for localizing hallucinated spans and enabling fine-grained interventions. In this paper, we explore the use of the Mixture-of-Experts (MoE) paradigm to address this gap. In MoE architectures, a single forward pass activates a sparse subset of experts (i.e., distinct feedforward networks per layer) via a routing mechanism, producing internal signals (e.g., router entropy, expert disagreement, and expert usage patterns) that are unavailable in dense architectures and have not been previously exploited for hallucination detection. To this end, we introduce InnerExpert, the first method to leverage these MoE-specific signals for per-token hallucination detection. InnerExpert combines routing-level and standard transformer signals into compact per-token feature vectors, classified by a lightweight detector trained on labels produced by an LLM-as-a-judge pipeline, which enables continuous model updates without manual annotation. Our results show that InnerExpert outperforms existing methods across five datasets and two MoE architectures, achieving up to 0.91 answer-level and 0.76 token-level AUROC, while requiring only a single forward pass.

Self-evolving agents turn experience into reusable skills, workflows, or memories, but post-evolution accuracy alone does not show whether learned behavior preserves previously correct behavior or security. We audit SkillOpt, Agent Workflow Memory (AWM), and ReasoningBank in simulated e-banking using matched benign acquisition trajectories, sealed evaluation endpoints, execution-grounded checks, and independent state replay. On Qwen 3.7 Flash, SkillOpt raises benign utility from 0.741 to 0.837 while exposure to injected content rises from 0.820 to 0.943. Conditional attack success after exposure falls from 0.605 to 0.562, yet overall attack success rate (ASR) rises from 0.496 to 0.530 and unauthorized financial state changes rise to 0.685. Across three independently evolved lineages, capability, exposure, and unauthorized-state changes increase in all three, whereas ASR increases in only two. ReasoningBank raises utility to 0.859 without increasing aggregate ASR, although unauthorized state changes remain slightly above Static. AWM reveals a separate evaluation hazard: a literal WebArena text-action envelope disrupts tool execution in our native function-calling executor. In a post-hoc sensitivity test, removing only that envelope restores utility from 0.319 to 0.756, while exposure rises from 0.299 to 0.909 and ASR from 0.195 to 0.575. Auditing self-evolving financial agents therefore requires tracking regressions, attack-surface contact, unauthorized financial-state change, and artifact-executor compatibility, not accuracy alone.

Drug discovery and development underpins healthcare but remains costly and failure-prone. A critical bottleneck lies in predicting molecular properties such as solubility, potency, and toxicity, which directly determine whether a candidate can advance from preclinical to clinical trials. Artificial Intelligence (AI) has accelerated this process, yet its reliability is often undermined by distribution shift, as experimental conditions frequently diverge from training data. In addition, conventional point predictions provide only single-value estimates, offering limited guidance for high-stakes experimental design. We address these challenges with a conformal prediction framework tailored to label shift. By weighting conformal scores using marginal label probability ratios, our method produces statistically rigorous prediction intervals without retraining. This enables robust uncertainty quantification even when property distributions drift, directly tackling one of the most pervasive obstacles to applying AI in real-world drug development. By moving beyond accuracy alone to provide actionable confidence measures, our approach enhances the trustworthiness of AI-driven predictions. This further aligns predictive modeling with regulatory demands for transparency and uncertainty reporting and ultimately supports more reliable decision-making in billion-dollar development pipelines.

Automated security patch backporting is critical for mitigating N-day vulnerabilities. Recent tools report success rates above 80% on their respective datasets. However, these evaluations are often confined to homogeneous environments, such as one repository or specific project versions. Consequently, it remains unclear how well these tools generalize beyond their originally targeted scenarios. We present Porting Benchmark, a curated dataset of 1,234 security patch backporting cases spanning cross-version, cross-branch, and cross-repository scenarios, paired with a common evaluation framework. Using this benchmark, we evaluate five tools spanning program analysis, LLM prompting, and LLM agents under aligned settings. Our results show that aligned evaluation changes the apparent performance landscape: PortGPT and TSBPort remain comparatively strong on the Replication Dataset, while FixMorph and Mystique degrade substantially under the common protocol. Performance degrades sharply on structurally complex patches: the best commit-level success rate falls from 85.2% on Type-I patches to 24.0% on Type-IV. We identify four root-cause categories (missing target API awareness, cross-version semantic mismatch, non-local dependency propagation failure, and patch construction or localization failure) and derive concrete directions for next-generation tool design. On a 45-case dynamically validated subset with verified test cases and constructed POCs, we further observe that reference-based benchmark scores do not fully capture real-world remediation: exact match sharply under-credits harder target adaptations, while executable validation reveals residual integration failures in the target that static reference agreement misses. Executable-feedback refinement provides limited but measurable recovery on the hardest executable cases.

Bayesian imaging inverse problems often require sampling from high-dimensional posterior distributions. While recent score-based and diffusion models provide expressive Bayesian priors, their sampling procedures remain inherently sequential and computationally expensive for large-scale imaging applications. We propose PiX-MC, a time-parallel posterior sampling framework based on proximal Langevin dynamics and Picard iteration. The proximal-likelihood formulation exploits the fact that many imaging likelihoods admit efficient, problem-specific proximal operators, while Picard refinement exposes parallelism across discretization nodes and naturally supports multi-GPU implementation. To further improve practical scalability and sampling performance, we develop multi-block and annealed variants of the proposed framework. We establish convergence guarantees under transparent assumptions, accommodating non-log-concave posteriors, imperfect learned score models, multi-block implementations, and annealing schedules. Experiments on a diverse collection of imaging inverse problems demonstrate that PiX-MC substantially reduces wall-clock time while preserving reconstruction quality. On a $512\times512\times80$ sparse-view computed tomography (CT) problem, annealed multi-block PiX-MC achieves up to a $50\times$ runtime speedup over the standard Langevin sampler using eight GPUs.

LLM-driven agents can autonomously exchange opinions on online platforms and form communities. Such agent-operated social platforms raise a new security concern: attackers may manipulate agents to induce group polarization. Existing methods manipulate agent prompts or construct echo chambers, both of which are difficult to realize in practice. We therefore formulate a new threat, Memory-Mediated Polarization Cascade, which uses agent memory as a persistence channel and public discussion as a propagation channel. This threat contains three stages. During exposure and memory retention, the attacker exposes a small set of target agents to arguments that reinforce their respective stated stances. The targets' memory systems then process and retain these arguments. During retrieval and reproduction, a shared stance-neutral discussion cues the targets to retrieve and reproduce their respective retained arguments. During iterative propagation, untreated agents influenced by the reproduced arguments restate and spread them. We instantiate this threat in GraphWake with three components: (i) stance-support argumentation knowledge graphs construct knowledge-based arguments; (ii) axiom-oriented triple selection distills them for reliable retention and reproduction; and (iii) stance-neutral memory cueing triggers concurrent retrieval and reproduction, initiating propagation. Experiments across multiple discussions and memory systems show that GraphWake substantially increases group polarization. These findings reveal a community-level polarization risk.

LLM-powered GUI agents that autonomously operate smartphones are rapidly transitioning from research prototypes to early real-world deployment. However, because these agents routinely process untrusted environmental content, they are highly vulnerable to environmental injection attacks, which include indirect prompt injections and adversarial instructions. Such attacks can manipulate the behavior of agents without user awareness through diverse channels encountered in everyday mobile use. Despite these risks, existing benchmarks often fail to capture everyday user scenarios, lacking a systematic evaluation of GUI agents under environmental injection attacks on mobile devices. To address this gap, we introduce MobileWorldSafety, a benchmark of 142 risk tasks built on real Android applications. For each task, we define a programmatically verifiable risk indicator over the final system state and evaluate outcomes with a two-stage pipeline: rule-based verification handles unambiguous cases, while an LLM judge adjudicates ambiguous ones. This distinguishes safety failures from capability failures and enables objective and reproducible assessment. Evaluations on six agents, including both general agents and specialized GUI agents, demonstrate that all agents remain highly vulnerable, with attack success rates ranging from 40.4% to 66.9%. These findings indicate that current agents often fail to maintain safety alignment when adversarial content is presented as ordinary mobile context. MobileWorldSafety provides a foundation for quantifying these vulnerabilities and advancing research on robust mobile GUI agents.

This monograph develops elimination geometry (EG), a typed, native-loss, audit-oriented framework for studying when locally optimal objects can be realized by a shared deployment rule. Elimination and compression may erase distinctions required by prediction, inference, control, or representation. EG asks which distinctions are lost, whether the induced defect is visible to the declared task, and whether changing information, architecture, action space, or deployment domain can repair it. EG separates local solvability, global realizability, and finite-sample certifiability. It derives native defects from the original objective and distinguishes architecture obstruction from model approximation, generalization, and implementation error. The monograph synthesizes tools from geometry, optimization, information theory, statistics, and machine learning into interfaces for integrability, representation admissibility, resource constraints, observational overlap, and common deployment. Formal results address regular, coordination, singular, compositional, and resource-limited mechanisms with explicit antecedents and claim boundaries. Applications include sparse model selection, distribution-free prediction, observational treatment policies, routed expert and retrieval systems, and learned score fields. Obstruction-Aware Learning and Inference links structural diagnosis to finite-data authorization, mechanism-matched intervention, and independent validation. Reproducible synthetic and real-data studies illustrate how certificates can guide architecture repair while recording failed gates and unresolved cases. The framework requires the deployment contract, native endpoint, competing explanations, information and compute budgets, and validation rule to be fixed before a persistent performance floor is attributed to architecture.

Agents increasingly interpret a person's natural-language preferences by querying an LLM for numerical preference judgments, e.g., by asking how much the person would be willing to pay for an item. A growing body of work estimates a utility function from these judgments and then chooses actions based on their estimated utility. This pipeline assumes the judgments are approximately self-consistent: that a single utility function can reproduce them. But are they? To study this question, we measure the self-consistency of cardinal LLM preference judgments. For example, the difference in stated willingness-to-pay between two items should match the stated payment that makes a person indifferent to exchanging them. We develop statistical tests and interpretable measures of how far observed responses depart from the best-fitting self-consistent utility function. Experiments with flight, apartment, and hotel examples across six LLMs reveal large persistent inconsistencies. This suggests that LLM-derived preference judgments cannot be faithfully summarized by a single utility function.

Modern embedded systems are becoming increasingly complex and typically integrate numerous heterogeneous devices, such as controllers, sensors, actuators, and supporting subsystems. As a result, their development and integration involve a wide variety of serial communication protocols, ranging from standardized solutions to partially standardized and fully project-defined formats. Efficient development of such systems increasingly depends on automation toolchains, which in turn require a clear, unified, and machine-processable formal basis. This paper proposes a unified, protocol-agnostic message model for explicit and deterministic description of serial messages. The model is based on formal definition of data types, atomic message elements (containers), and complete message structure. In addition to the model itself, the paper introduces methods for practical work with it, including configurable message types for expressing structural constraints and supporting deterministic automation, as well as configurable user representations for engineering-oriented reading and editing. The proposed model and methods are demonstrated through implementation in an industrial tool environment. The results show that the approach can support machine-readable interface control document development, automated generation of transport-layer software, and practical engineering work with both standardized and weakly formalized serial protocols. Taken together, the proposed model, methods, and tool implementation provide a practical foundation for automation toolchains in heterogeneous serial communication development.

We present rl-triton, an open-source library of high-performance GPU kernels for reinforcement learning credit assignment, implemented in Triton. The core contribution is a unified associative scan framework that recasts seven distinct RL estimation algorithms - Generalized Advantage Estimation (GAE), V-Trace, Retrace($λ$), TD($λ$) returns, discounted returns, eligibility traces, and episodic prefix sums - as instances of a single first-order linear recurrence solved in $O(\log T)$ parallel steps. All algorithms share the same associative scan operator, with algorithm-specific fused Triton kernels constructing their recurrence coefficients on-chip. We verify the associative operator algebraically and define the treatment of terminated and truncated episodes explicitly. Benchmarks show a 1.6-5.70$\times$ full-call speedup over a vectorized torch.compile baseline in the massively parallel simulation regime (thousands of environments, short rollouts). The reported range covers all seven algorithms on both GPUs, both with and without per-step truncation handling. For most algorithms, speedups increase at longer sequence lengths, as the baseline requires more scan stages as $\log T$ grows, each adding an intermediate HBM round-trip. The library is available at https://github.com/simonsays1980/rl-triton.

What a reasoning model writes is only a partial record of the process that produces it. We introduce a two-level internal readout for mixture-of-experts reasoning. We first distill vocabulary-scale J-space into J64, a 64-axis semantic frame learned from the model's own reasoning states. J64 reveals readable process state that the emitted trace does not show: it separates inference effort from problem-induced strain. It also adds 0.096 to 0.135 held-out AUC over a baseline that reads the same rollout as token occupancy and aggregates it in exactly the same way. We then reconstruct J64 from native expert-routing statistics. The result is R64, a low-overhead proxy: its median per-axis correlation with J64 is 0.69 to 0.86 across three models and two families, and on gpt-oss-20b it preserves 95 to 100% of J64's predictive gain. The readout supports test-time decisions at two temporal resolutions. Over completed candidate sets, J64 and R64 improve single-branch selection, and R64-weighted voting improves plain majority voting in seven of eight settings. During generation, rolling readout windows drive a cumulative stop-and-resample policy whose operating point is fixed on training questions alone. J64 improves accuracy by 1.1 to 5.9 points over a sibling-permuted control, and the routing-only R64 proxy retains 0.9 to 3.2 of those points. Finally, router edits aimed at the mechanism J64 names induce the predicted reasoning behaviors and shift a diagnosed stall from numerical guessing toward exact symbolic execution. Together, J64 makes latent process state readable, while routing makes it deployable and actionable.

The $\operatorname{do}$-operator is described graphically by deleting arrows into its targets and functionally by replacing their mechanisms with constants. To call these operations equivalent is not yet a mathematical statement: one returns a graph and remembers only the targets, whereas the other returns mechanisms and also remembers the imposed values. We make a dependency-level comparison precise for deterministic acyclic structural causal models with finitely many endogenous variables. If $\operatorname{Graph}(F)$ extracts the dependencies of a mechanism family $F$, our main theorem is $\operatorname{Graph}(F^ι)=\operatorname{Surg}(\operatorname{Graph}(F),T_ι)$. Thus replacing target mechanisms removes exactly the dependencies removed by graph surgery. For a model $M=(G,F)$ whose graph may contain unused arrows, we characterize when the same equality holds with $G$ in place of $\operatorname{Graph}(F)$; it holds for every intervention exactly when $G$ records the dependencies of $F$ exactly. We then define the intervened model, characterize its run, show how sequential interventions combine, and prove that an outcome depends only on interventions at its actual dependency ancestors.

Large language models (LLMs) can both expand underspecified queries and encode text as dense representations, suggesting a unified model for query expansion and retrieval. Existing systems usually rely on prompted expansions, independently trained modules, or staged optimization, leaving generated expansions only indirectly aligned with the retrieval loss that judges them. We train a single decoder-only LLM end to end, where the same model generates the expansion and encodes both the expanded query and candidate documents. This unified setting creates a moving-target problem: retrieval supervision should improve query-side expansion, but the same update also shifts the document embeddings that serve as retrieval targets. We introduce Document Embedding Preservation Tuning (DEPT), which keeps tuned document embeddings close to cached initial embeddings while allowing retrieval gradients to pass through straight-through decoding into the generator. DEPT converts joint query--document movement into query-side adaptation against approximately stable, whitened document embeddings that support index reuse and online hard-negative mining. Experiments with Qwen3-4B-Instruct-2507 and LLaMA-3.2-3B-Instruct on five datasets in BEIR benchmark show that DEPT improves average retrieval quality over training-free, independently trained, and staged unified baselines, while ablations isolate the effects of preservation, whitening, end-to-end expansion training, and online negatives. Code is available at https://github.com/ILSparkle/DEPT.

Developing robots capable of understanding and manipulating objects requires compact, interpretable, and generalizable representations. This work proposes a reinforcement learning-based framework for robotic grasp refinement, integrating keypoint-based object representations with a Deep Q-Network (DQN). Using 2D overhead images captured in a simulated environment, a geometric-based algorithm generates initial grasp candidates, which are iteratively refined by the proposed framework, transforming failed grasps into successful ones. Experiments conducted on 300 objects from the Dex-Net dataset using a UR5 manipulator demonstrate the framework's effectiveness, achieving a 100% success rate on objects previously deemed ungraspable by geometrical methods. The framework's sim-to-real transferability is further validated through physical experiments on a Delta parallel robot, where a refined grasp successfully manipulates an object that was previously ungraspable. The findings underscore the effectiveness of reinforcement learning in addressing challenges in robotic grasping, offering a scalable and adaptable solution for contact-rich manipulation tasks.

Saudi Arabia will host the 2034 FIFA World Cup and already operates crowd management at Hajj scale. Drone-based counting must hold accuracy on footage unlike anything in its training corpus, without labels, and must warn of dangerous inflow before a crush forms. We deliver a validated answer built on 525 controlled runs, a full-resolution corpus study, five falsification ablations, and a five-condition safety-interlock evaluation. Label-free adaptation recovers 31-49% of shift-induced error across four corruptions and five severities, with the strongest method gaining 41.8 MAE over the frozen source (95% CI [34.1, 49.6], p=7.5x10^-10, d=2.52). We establish a severity law separating methods with a constant absolute margin from the one whose margin grows, and a stability budget identifying which configuration is safe to fly. On a full-resolution corpus carrying a genuine +48 MAE aerial gap (source retrained to 14.6 validation MAE, a 34% improvement), adaptation repairs the dense-scene undercounting that would otherwise under-report a forming crush, and the flux-based risk module fires on real congestion episodes in 2 of 6 full-length clips. We localise the recoverable error: in a regime built to favor a physics-informed conservation prior (300-frame clips at 200ms spacing, five times wider than standard), the adaptation signal is normalisation-driven, not flow-driven; the continuity residual is invariant to the proportional counting errors domain shift produces, confirmed by four on/off ablations correlated at r=0.999 and a 40% input corruption moving accuracy by only 0.05 MAE. A label-free shift gate shows shift magnitude and accuracy damage are rank-independent (Spearman rho=0.20; rho=-0.60 among genuine shifts), quantifying the 58% of headroom a magnitude gate forgoes. We establish unconditional adaptation with tail monitoring as policy, closing with a six-point protocol.

Machine learning models for electroencephalography (EEG) analysis show great promise across a wide range of applications, but their deployment in high-risk domains is hindered by their vulnerability to distribution shifts. Encountering out-of-distribution (OOD) data can lead to catastrophic, overconfident predictive failures. While OOD detection methods can mitigate these risks, they remain heavily under-explored for EEG. Moreover, evaluations in the broader literature typically evaluate OOD detection performance in isolation, ignoring their practical impact on downstream applications. To bridge this gap, we introduce a benchmark for EEG OOD detection, evaluate a broad range of methods, and furthermore evaluate their value in two clinical downstream prediction task. Our results disentangle OOD detection and model uncertainty estimation capabilities, which are frequently conflated in the literature, provide actionable insights about the current state of the art for EEG OOD detection and model uncertainty estimation, and demonstrate how complementary methods for both aspects can be combined to form a robust safety net for the deployment of EEG-based machine learning models in real-world applications.

Learning analytics models can identify students at risk of poor performance, but they do not directly indicate which interventions are feasible, actionable, and compatible with educational constraints. This paper introduces SC2R, a semantics-constrained counterfactual recourse framework for educational decision support. SC2R combines a calibrated predictive model, integer-programming-based recourse generation over discrete action variables, a lightweight RDF vocabulary for intervention-plan representation, and SHACL validation for enforcing timing, budget, immutability, and availability constraints. The framework is evaluated offline on the OULAD dataset using snapshots constructed relative to each assessment at two decision horizons. Results show that the predictive component provides strong performance, that compact intervention plans can be generated at scale, and that semantic validation reveals infeasible plans that lighter optimization-only settings would otherwise accept. Rather than claiming causal improvement in student outcomes, this work shows that counterfactual recourse becomes more operationally meaningful in education when recommendations are not only model-valid, but also semantically feasible and machine-checkable.

We present MoNe, a lightweight modular neural memory that attaches to any frozen pretrained Transformer to enable long-context inference without retraining. MoNe reads context in fixed-size segments via test-time learning of fast-weight neural memory networks with layer-localized gradient updates; at inference, the memory generates keys and values from the query tokens alone, with no context tokens re-read. This two-phase design decouples inference cost from context length, achieving $O(N)$ preprocessing and $O(1)$ query cost with peak GPU memory that does not grow with $N$. At 128K tokens, MoNe reduces both compute and peak GPU memory by approximately 80% compared to ICL with only 6.4% parameter overhead. MoNe generalizes to context lengths far beyond the backbone's native window, achieving strong performance on needle-in-a-haystack and word extraction benchmarks from RULER, where ICL degrades sharply.

Conversational AI is moving beyond isolated text prompts toward sustained, multimodal interaction. In real conversations, users clarify goals, revise requests, interrupt responses, switch topics, and introduce new evidence while expecting systems to preserve context across turns. This makes multi-turn dialogue a distinct challenge requiring systems to maintain and update memory, ground responses across modalities, tools, and external knowledge, and adapt across languages and cultures. This study reviews multi-turn conversational AI across text-only dialogue, AudioLLMs and speech-native systems, multimodal and omni-modal systems, and tool-augmented agents. We organize the literature around datasets and benchmarks, modeling paradigms, training strategies, evaluation setups, and cross-cutting challenges. Our analysis shows that support for multiple modalities has advanced faster than the ability to sustain coherent interaction across a session. Despite stronger capabilities to perceive, speak, and act across modalities, current systems still struggle with persistent memory, cross-turn grounding, full-duplex interaction, robust evaluation, and cultural alignment. We conclude with a research agenda for systems that can remember, revise, ground, speak, listen, act, and adapt across turns, modalities, and cultures. (https://github.com/faiza-sfa/multiturn-conversational-ai-survey)

Large language models are increasingly deployed through agent harnesses that manage tools, extensions, persistent state, permissions, and external actions. Existing safety benchmarks mainly target individual attack mechanisms or a limited subset of operational settings, making it difficult to compare how safety failures emerge across different harness responsibilities. We present HarnessRisk, a lifecycle oriented benchmark that organizes agent harness safety into six operational phases including Harness Configuration, Capability Extension, Runtime Operation, State Persistence, Action Control, and Incident Recovery. HarnessRisk contains 128 sandboxed cases, each pairing a benign user objective with an adversarial instruction embedded in an untrusted workflow artifact. We evaluate each trajectory using Utility, Attack Success Rate, Persistence, and Detection. Across three harnesses, six language models, and 14 model and harness configurations, attack success ranges from 12.6% to 80.9%, while Utility remains between 75.0% and 97.6%. Harness Configuration is the most vulnerable phase across all three harnesses, showing that attacks can succeed by altering security sensitive parameters within otherwise authorized workflows. We also find that explicit risk recognition does not reliably lead to safe action, as some configurations detect risks in more than 90% of runs while retaining substantial attack success. These results highlight the need to evaluate agent safety across multiple harness responsibilities and at the level of the deployed model and harness configuration.

In this study, we investigate developmental mechanisms that enable small, resource-constrained systems such as cm-sized millirobots to autonomously explore, learn, and adapt their capabilities throughout their lifespan. Reinforcement learning algorithms guide the agent's skill acquisition and adaptation through the interplay of our proposed tinyDSM, which integrates intrinsic motivation and fitness-based assessment. We strive for minimal, hard-wired skills while encouraging the open-ended development of new skills. A key emphasis in our approach is to encode minimal a-priori general knowledge, which serves as a foundational starting point for the system as it further learns system-specific dependencies from the initial knowledge provided. Thus, by design, our approach attempts to cover very generic application domains. The methodology is based on (a) developmental mechanism with intrinsic motivation, and (b) a cognitive architecture (knowledge, reasoning, learning), while (c) utilizing minimal resources. It uses a hierarchical knowledge graph and kinematic reasoners to model and evaluate simple and advanced motion related skills. In our experiments, we use a resource-constrained millirobot with a volume of 36 cm^3 with a Raspberry Pi Pico 32-bit microcontroller (RP2040) that integrates all described features and capabilities except the camera system in 9 kB. Starting with learning the most elementary motor skills the millirobot autonomously progresses from simple linear and angular movements to complex geometric patterns within 15 minutes. To complement the physical experiments, we perform a simulation-based analysis that enables systematic comparisons across learning algorithms and intrinsic motivation parameters.

The communication demands of distributed model prediction control (DMPC) can overwhelm even advanced wireless communication technologies as agents must exchange a significant amount of information at least once per time step. To semantically reduce communication demands, this work employs encoder-decoder networks built around long-short term memory (LSTM) cells in a distributed optimization algorithm. Agents publish a reduced representation of a message and receivers reconstruct the original message upon reception. In tests with reduced communication using formations of mobile robots, trained networks retain satisfactory performance and work reliably under conditions overwhelming full communication. As the results show, the usage of LSTMs either allows unprecedented reconstruction accuracy or the usage of different prediction-horizon lengths without the necessity to retrain.

Agent Skills package reusable natural language procedures with executable resources, enabling software agents to acquire task specific capabilities without model adaptation. Automatically generating such Skills can improve task performance, yet evaluating a candidate solely from its artifact or final task outcome leaves unresolved which actions the equipped agent will perform and which side effects those actions will produce. We present TRUSS, an evidence guided framework for generating functionally effective and safety reliable Agent Skills. TRUSS first inspects functional claims against source and domain evidence while evaluating the complete artifact under nine predefined safety properties. Candidates admitted by this static gate are loaded by a shadow agent inside a Controllable Execution Environment, where brokered tools expose requested actions to policy enforcement and record their results as provenance preserving execution traces. Functional failures and property violations are linked back to the responsible Skill content and used to guide iterative refinement. We evaluate TRUSS on 168 SkillInject artifacts, 155 SkillSafetyBench cases, and all 187 tasks in SkillGenBench. TRUSS achieves 100.00\% precision and recall in vulnerability detection. Repair reduces attack success from 38.71\% to 19.35\% with GPT 5.5 and from 46.45\% to 29.68\% with GPT 5.4, with zero attack regression. For Skill generation, TRUSS raises task effectiveness from 17.11\% without Skills to 52.94\%, while increasing the benchmark Security rate from 50.80\% to 100.00\%. These results show that execution evidence can expose behavioral failures missed by artifact inspection and can guide Skill generation toward jointly verified functional and safety outcomes.

Expert-written natural language skills can improve tool-using agents, yet agent-authored skills perform 8-11 points worse than using no skill. This gap suggests that following procedural guidance and improving it from execution evidence are distinct capabilities. Inference time loops can repair skills but do not improve the model that writes the next one. We study how to organize execution experience from intermediate skills into training states for an optimizer. We introduce WER (Write, Execute, and Refine), a multi-phase framework that trains a Skill Optimizer outside a frozen executor. The optimizer proposes skills, a frozen agent executes each repeatedly, and a programmatic verifier scores the outcomes. The scores provide relative credit and select mixed-outcome records. Matched successful and failed trajectories from these records form the next phase's refinement states, so the optimizer learns from the consequences of its earlier outputs. On BFCL v4 multi-turn and tau2-bench, WER improves average Pass@1 over the no-skill baseline by 7.80 and 3.85 points, respectively. Under an identical refinement workflow, it outperforms the same backbone without optimizer training by 9.35 and 10.29 points. The trained 4B optimizer reaches 76.63 percent on BFCL v4, outperforming all evaluated off-the-shelf general-purpose models used as skill optimizers on average.

Online video platforms can expose young users to harmful content, but independent audits remain difficult because video annotation is costly and moderation judgments vary across languages. We audit TikTok in France, Italy, and Sweden with sockpuppet accounts representing four age personas (13, 16, 19, 40), collecting 36,971 videos from passive For-You-page scrolling and active sessions that scroll, search for harm keywords, and scroll again. To scale annotation, we validate four multimodal LLMs against native-speaker labels on a 300-video reference set. Gemini 2.5 Flash with eight sampled frames plus text performs best (aggregate kappa = 0.42), at half the per-call cost of native-video upload, and we apply it to a 10% sample for approximately \$50 in total API spend across both modalities. Keyword search returns 35-56% harmful content, a 1.5-7.5x increase over the scrolling baseline in ten of twelve country-age combinations; the spike is temporary and flattens the age differences observed in France and Sweden. Under passive scrolling, Italy has the highest harm rate at every age, with Italian age-19 reaching 48.6%. Overall, MLLM-based auditing offers a scalable approach for cross-national youth-safety audits, while provider safety filters (1.1% refusal rate) under-count the most explicit harms.

How cautious should an agent be while it is still learning its environment? We propose RATTL (Risk-Adversarial Total-Reward Learning), which ties caution to epistemic uncertainty: the agent holds a Bayesian posterior over unknown dynamics and plans against a Wasserstein ambiguity set whose radius is a monotone function of that posterior. The radius contracts with evidence, so behaviour interpolates continuously between worst-case robustness and risk-neutral total-reward maximization. The design follows the duality underlying the Entropic Value-at-Risk, which converts the choice of a risk level into the choice of an ambiguity radius. We show the resulting planning problem is well posed under transience and compactness conditions, and prove a Safety Sandwich: the RATTL value lies between the uninformed robust value and the full- knowledge optimum, with a gap that vanishes as the posterior concentrates. In a canonical binary-hazard instance, the induced criterion reduces to Conditional Value-at-Risk at a level set by the posterior entropy. A worked example shows the agent deferring the efficient action until a sharp identification threshold. RATTL targets runtime safety for agents, including LLM-based systems, acting under uncertainty.

In high-dimensional online prediction, the best predictor may depend on only a few features, so regret should scale with sparsity rather than the ambient dimension. Feature priming pursues this goal by estimating feature weights from past data and refitting a minimum-norm predictor on the rescaled design. Warmuth and Amid asked at COLT 2023 whether any of three such rules admits a competitive online regret guarantee. Using the natural Moore--Penrose protocol based only on past data, we give a negative answer to the sparse-logarithmic form of this COLT open problem. Our analysis identifies a common obstruction: cheap nuisance interpolation causes the refit to underweight the truly predictive coordinate. An exact target-mass identity and a two-sign argument turn this effect into clipped prediction loss. Hadamard constructions force $Ω(\min\{T,\sqrt{d}\})$ regret for all three rules against a zero-loss one-sparse comparator, with extensions to fixed prime powers and selectors among the rules. Conversely, regret is controlled by data rank, and a Euclidean-normalized triangular construction matches this dependence for powered univariate priming, even under nonnegative second-stage ridge regularization; a paired ridge construction also covers all three powered rules. Exploratory diagnostics on frozen language-model activations exhibit the same relation among nuisance interpolation, target weight, and loss. The exact multivariate and Pearson frontiers remain open.

Motivation: Low-dimensional embeddings are widely used to explore cell-state heterogeneity in single-cell and other high-dimensional biological data. Although many methods preserve local neighborhoods, they may distort the apparent sampling density of processed observations, altering the visual contrast between dense and sparse regions and complicating the interpretation of rare, transitional, or continuous cell-state populations. Results: We present DMT-Dens, a parametric manifold-visualization method built on a latent-token Transformer encoder. The model integrates rank-based manifold alignment with hard-pair aggregation. To preserve density, it optimizes a loss based on the Pearson correlation between k-nearest-neighbor log-radius estimates in the processed input and two-dimensional embedding spaces. Benchmark evaluations demonstrate strong density preservation, particularly on biological datasets, while retaining competitive label separability. Availability: Source code, data-processing scripts, and resolved experiment configurations are available at https://github.com/Ruizhe-wang/DMT-Dens.

Pretrained molecular language models are increasingly used as molecular encoders for learning structure-property relationships. However, their practical suitability for molecular discovery within and beyond their pretraining domain remains unclear. Herein, we systematically benchmark four molecular language models across six virtual molecular libraries spanning drug discovery, organic materials, and catalysis. Native molecular language model embeddings show substantial variation in discovery performance across libraries, whereas molecular fingerprints provide a consistently strong and robust baseline. Consistent with a potential domain-representation mismatch, we show that explicit domain adaptation substantially improves representation performance. Fine-tuning molecular language model encoders on structures from the target virtual library consistently improves sample efficiency, with several adapted encoders emerging as the top-performing representations across the benchmark tasks. These results show that molecular representation quality depends strongly on the target domain and that explicit adaptation can improve the practical utility of molecular foundation models. More broadly, our findings establish domain-adapted molecular representations as a promising strategy for sample-efficient adaptive decision making in virtual screening and self-driving laboratories.

Unified multimodal models (UMMs) are motivated by the hope that understanding and generation reinforce each other but controlled ablations repeatedly find that adding a generation objective leaves understanding flat. Joint-training studies cannot settle the disagreement: with overlapping supervision, a gain cannot be attributed to the architecture rather than the data. To further investigate the relationship between the two directions in UMMs, we separate them by construction. A novel visual entity, a rendered 3D asset paired with a pseudo-word screened for absence from the frozen model's behavior, is bound through exactly one task direction, and the untrained direction is then measured. We find that the channel is real in both directions, but the directions differ in kind: generation training installs a name the model can only match among candidates; understanding training installs one it can also produce. What governs cross-task usability is where the binding enters the shared computation. An alignment probe predicts export across 36 configurations (Spearman $ρ= +0.68$). That objective's alignment term, maximized in closed form over activations with every weight frozen, makes a concept drawable when injected at layer 7 of 28 and is indistinguishable from the base model from layer 14 on, while the weight-based version of the same edit peaks at layers 10-14. In an observational series of four models, this window appears only where the understanding pathway is a semantic vision encoder, suggesting that unified weights are not enough: the two directions must share a semantic format at the entry point. Exploiting the rule, a mid-stack alignment objective acquires the concept for a $0.1\%$ relative loss of the model's general text-to-image ability, against $41\%$ for the standard generative route. Our code is at https://github.com/Zane-ZYQiu/entry-point-umm.

The health of marine ecosystems is a critical indicator of global environmental change, yet the physical constraints of underwater observation and the intrinsic challenges of processing marine imagery severely limit the scalability of systematic monitoring. While recent visual foundation models such as the Segment Anything Model (SAM) series show great promise, they still struggle with the fine-grained recognition required in these complex scenarios and still require expert supervision. Our work addresses this gap by bridging state-of-the-art foundation models with existing sparse supervision. Because historical benthic surveys are typically annotated with only a few sparse expert points per image, we utilize these legacy point-labels as visual prompts for SAM2. Our primary contribution is a novel mechanism to automatically identify which of these points are suitable, and which are actively harmful, when used for propagation. By filtering out unreliable points, we extract high-quality pseudo-ground-truth masks capable of training more accurate, fine-grained semantic segmentation models. We demonstrate the effectiveness of our approach on public benthic data and introduce a new, challenging benchmark featuring real-world sparse expert annotations, paving the way for scalable ecological analysis.

Large Language Models (LLMs) in real-world applications often face the risks of specially crafted prompts designed to bypass the safety controls. Existing guardrail methods, such as LLM-as-a-judge and cloud-based safety APIs are able to detect unsafe content. However, they often add a delay of about 250-900 ms to each request. This delay is too high for real-time applications, when the system usually needs to respond in less than 100 ms. Furthermore, routing user prompts through external moderation endpoints raises significant data privacy concerns. This paper introduces Reflex-Guard, a lightweight guardrail that runs locally. It uses jailbreak-aware preprocessing, compact sentence-transformer embeddings, and seven fast binary classifiers. Together, these components enable high-accuracy prompt safety filtering with much lower latency than existing solutions. Through systematic evaluation on a strategically balanced dataset of 30,568 samples drawn from five complementary sources, we demonstrate that Reflex-Guard achieves 95.9% recall on harmful prompts at 37.6 ms end-to-end latency. It is faster than existing baselines, including Llama Guard 2 at 255 ms and SafeDecoding at 723 ms. It can detect 100% of GCG suffix attacks and Base64-encoded prompts using the default threshold. However, DrAttack structured prompts required lowering the threshold to 0.03 for optimal detection, as they produced a distinct probability distribution. Reflex-Guard achieves Reflex Efficiency Score (RES) scores up to 16.79, significantly outperforming Llama Guard 2 (11.90) and SafeDecoding (9.80). This analysis offers practical deployment advice and shows that different attack types occupy distinct regions in the embedding probability space.

Academic papers are a primary carrier of scientific knowledge, yet most of this knowledge remains locked in PDFs that are optimized for human reading rather than machine use. For Multimodal Large Language Models (MLLMs), the core challenge is not only perception, but representation: scientific pages interleave text with Structured Academic Elements (SAEs) such as tables, formulas, charts, and pseudocode, whose structure, data, and logic are poorly preserved by common surrogates like Markdown. We therefore propose Compilable Academic Document Parsing (CADP), a paradigm that reconstructs a full page as contextual \LaTeX{} plus executable Python, so that structure-preserving elements and executable chart representations can be reconstructed, recompiled, and directly verified against the source page. To support this setting, we introduce CADP-Bench, an expert-verified benchmark of full academic pages containing tightly coupled text and multiple SAE types, evaluated through a re-injection compilation protocol. We further study current capabilities using SOTA MLLMs and an exploratory multi-agent baseline that incorporates common agentic techniques. Results show that even frontier models still struggle to produce high-fidelity executable reconstructions, highlighting substantial room for improvement in structure-aware scientific document parsing. CADP-Bench is released for future research.

Testing RESTful APIs requires generating sequences of API calls that satisfy dependencies among operations, parameters, and runtime-created resources. Recent LLM-based approaches infer such dependencies and generate test sequences from OpenAPI specifications, but they often treat LLM-inferred relationships as correct without execution-based validation. This can introduce spurious dependencies, miss feasible operation chains, and produce infeasible tests. In this paper, we propose APIPilot}, an execution-validated framework for REST API testing. APIPilot first derives candidate producer-consumer dependencies from OpenAPI specifications using structural heuristics and LLM-based semantic reasoning. It then treats these dependencies as hypotheses and validates them through concrete API executions before using them for test generation. The validated dependencies are organized into a dependency graph from which APIPilot constructs coverage-aware workflows via bounded top-k graph traversal, separating semantic dependency inference from sequence construction. To improve subsequent tests, APIPilot further performs response-driven refinement: runtime responses are analyzed to update resource pools, adjust input-generation constraints, and prune or revise invalid dependency mappings. Empirical evaluation on 16 real-world REST API services shows that APIPilot achieves 92.3% operation coverage, up to 58.6% code coverage, and an 88.1% workflow execution success rate, outperforming both LLM-based and traditional REST API testing baselines. APIPilot also detects 197 unique 5xx failures and specification-execution mismatches, demonstrating the benefit of grounding dependency inference in execution feedback.

Joint-Embedding Predictive Architectures (JEPAs) learn world models by predicting future embeddings, but the objective admits a trivial solution of a constant encoder, so every practical system adds an anti-collapse mechanism (LeCun, 2022; Assran et al., 2023; Bardes et al., 2022; 2024). LeWorldModel (LeWM) prevents collapse with SIGReg, a regularizer that forces the latent distribution to match an isotropic Gaussian: the representation is stabilized by prescribing what it must look like, independently of the environment it models. We argue that the anti-collapse pressure can instead come from the transition data itself. Action-Contrastive Masked Transition Modeling (AC-MTM) keeps LeWM's forward latent-prediction objective and adds a training-only inverse-dynamics head trained with Action-NCE: each latent transition must identify the action that produced it among the other actions in the batch, a discrimination task that a collapsed encoder provably fails. The inverse branch is discarded after training, leaving test-time encoding, forward prediction, planning, and compute identical to LeWM. On four standard pixel-control tasks under a matched planning protocol, AC-MTM trains stably from scratch and matches SIGReg on average. On the harder multi-object OGBench Visual Scene task, results are consistent with the prescribed geometry becoming a bottleneck: AC-MTM reaches 80.0$\pm$2.0% success versus 58.0$\pm$2.0% for SIGReg, improving by 20-24 points in each training seed. A single 50-episode random-policy run gives a 52% baseline estimate. Contrastive inverse dynamics thus provides a distribution-free anti-collapse signal that requires no target network, stop-gradient, pretrained encoder, or reconstruction objective, and we characterize the action-space and observability assumptions under which it holds. We make our code available at https://github.com/jackboyla/action-contrastive-jepa

Legal consultation questions exhibit multi-level complexity. A single retrieval strategy often leads to over-reasoning for simple questions and poor interpretability for complex ones, making it difficult to meet the requirements for both answer quality and efficiency in high-risk scenarios. To address this issue, this paper proposes CoAL-RAG, a complexity-aware legal retrieval-augmented generation method, which constructs a multi-dimensional evaluation mechanism based on ``question essence'' and ``retrieval consistency'' to enable adaptive routing of retrieval strategies. First, the reasoning demand is quantified according to the logical structure of the question. Then, the discrepancy between semantic retrieval and keyword retrieval is utilized to indirectly reflect problem complexity, thereby selecting the most appropriate retrieval strategy and dynamically filtering contextual information. Experimental results demonstrate that the proposed method significantly outperforms baseline models not only on Chinese legal benchmarks (SocialLawQA, LawBench) but also demonstrates strong cross-jurisdictional generalization on English datasets (LexGLUE, CaseHold). Specifically, on Chinese datasets, the BLEU score improves by 42.5\% and ROUGE-L reaches 3.6 times that of knowledge graph-based methods. On English benchmarks, CoAL-RAG maintains highly competitive accuracy, achieving an optimal balance between generation quality, deep logical reasoning, and system efficiency across different legal systems.

Large language models increasingly serve as persistent conversational assistants, requiring memory that preserves relevant experience and maintains continuity across interactions. Existing methods improve access to conversational history through long-context processing, selective retrieval, and structured memory organization. However, most systems treat memory access as retrieving relevant past information without first determining which prior interaction state the current turn resumes. This limitation becomes particularly important when conversations interleave multiple tasks, people, and plans that may be interrupted and later revisited. We introduce ArborMem, an online memory framework that represents a long-running conversation as a navigable forest of interaction states. Each branch preserves a locally coherent trajectory, while the forest maintains multiple trajectories that may later be resumed. For each new input, ArborMem localizes the relevant state, restores its branch-local context, and augments it with reusable evidence retrieved across branches, preserving interaction continuity without conflating semantically related but structurally distinct trajectories. Existing long-term memory benchmarks cover diverse memory and reasoning capabilities but do not explicitly isolate branch-structured challenges. We therefore introduce BranchMemEval, a controlled diagnostic benchmark for interleaved and resumable interaction trajectories. Experiments on LongMemEval, LoCoMo, BEAM 100K, and BranchMemEval show that ArborMem outperforms the strongest baselines by 3.36 to 10.31 percentage points on the three established benchmarks and by 5.0 points on BranchMemEval. Its advantage grows under constrained read budgets, while complete memory queries remain below half a second.

Continual pre-training of large language models must acquire new information without erasing old knowledge. Existing replay methods often choose a global old/new mixture and sample uniformly, ignoring that examples differ in how quickly they are forgotten. We formulate continual pre-training as adaptive review scheduling: the training loop should decide not only how much history to replay, but which examples should return at each step. We introduce Spaced Repetition Training (SRT), a continual learning framework inspired by cognitive science, which schedules sample-rehearsal using the SuperMemo-2 (SM-2) algorithm. SRT maintains per-example review state, maps per-example perplexity to a recall-quality signal, and schedules historical examples for retention and new examples for consolidation while leaving the model, objective, and optimizer unchanged. On temporally separated Wikipedia and code corpora, SRT improves the stability-plasticity trade-off, recovering 5 to 37 percentage points of old-knowledge accuracy lost by naive continual pre-training across model scales while preserving or improving new-knowledge acquisition. At larger scale, SRT preserves broad benchmark performance that naive continual pre-training and uniform replay substantially degrade. Experiments with vision and tabular data further suggest that the scheduling principle extends beyond language when paired with an appropriate recall signal.

Modern agents operate inside agent harnesses that manage tools, context, and control flow, making the harness a critical part of the agent system. Our original Agent Lightning introduced a disaggregated architecture that connects arbitrary agents to RL training through an LLM endpoint proxy, an approach later adopted by frameworks such as verl Uni-Agent, AReaL 2.0, slime, and Polar. We refer to this paradigm as harnessed agentic RL, where the deploy-time harness directly participates in model post-training. Harnessed agentic RL differs fundamentally from traditional agentic RL: the harness, rather than the training engine, owns the environment interaction loop, while the trainer observes only sequences of LLM request-response pairs. This introduces challenges in retokenization, sample merging, advantage calculation, loss normalization, and backend scheduling, which can substantially affect training stability and effectiveness. We present Agent Lightning v1.0, a lightweight framework for harnessed agentic RL implemented in approximately 3,500 lines of code. It supports arbitrary agent harnesses and serves as a practical testbed for studying these challenges. We evaluate it on instruction-following, search, and coding agents, and provide a complete reproducible pipeline for coding-agent RL. Using only 6K training examples and modest compute, RL improves Qwen3.5-9B on SWE-bench Verified from 41.8% to 56.4%, a 14.6-point absolute gain. We release the complete workflow and training scripts to facilitate reproducible research on harnessed agentic RL.

This preliminary paper outlines a planned evaluation benchmark for Explainable Reinforcement Learning (XRL) methods. Current evaluations rely on functionally-grounded metrics like faithfulness and compactness, and on human-grounded proxies like subjective ratings or prediction accuracy. We suggest evaluating XRL methods by how effectively their generated explanations help to diagnose and fix malfunctioning reinforcement learning (RL) agents. We propose EvalXRL, a benchmark in which a Large Language Model (LLM) coding agent uses different XRL methods to diagnose a held-out malfunction in an RL agent, and then repair it. Our proposed benchmark iterates across (environment $\times$ malfunction $\times$ XRL method) tuples and uses the reward signal of the RL agents to form a final score for each XRL method. The coding agent may use the method interactively: invoke the XRL method, process its output, form new hypotheses on what is broken, and invoke the method again with parameters adjusted for testing these hypotheses. This closed-loop structure may be described as a simplified version of the scientific method. Some XRL methods provide self-evaluations that follow this pattern; we propose the first head-to-head comparison of multiple XRL methods in closed-loop usage.

Persistent shortages in the surgical workforce and inherent limitations of traditional training methods highlight the necessity of automated, data-driven approaches in surgical education. This study addresses these challenges by introducing a novel, explainable AI-powered framework for automated skill assessment, specifically focusing on cataract surgery. We present the world's largest dataset of cataract surgery videos, comprising 2,000 recordings. Additionally, we propose an AI-powered analytical framework that employs advanced computer vision and signal-processing techniques to automatically evaluate surgical videos to derive objective, quantitative performance indicators that complement or potentially replace subjective scoring methods. A significant advantage of our framework over previous methods lies precisely in its explainability of outputs, elevating it beyond merely an opaque skill classification tool. Through experimental analysis of 83 cataract surgery videos, we demonstrate that the automatically computed metrics exhibit strong correlations with expert-based subjective evaluations, achieving up to 87% accuracy in surgical skill assessment. Each metric was individually examined, and expert surgeons provided subjective ratings using the newly introduced Capsulorhexis Skill Assessment System (CSAS). These subjective assessments were compared with ten objective motion-based metrics extracted through our framework. The results indicated a robust correlation between subjective ratings and automated indicators, underscoring the framework's capacity to accurately model surgical expertise.

Vision-based surgical skill assessment has shown strong in-domain results, yet a fundamental question remains unasked: do these models learn transferable representations of surgical proficiency, or do they merely encode dataset-specific visual patterns? This paper systematically analyzes what limits cross-domain skill transfer between the GOALS and OSATS assessment scales using the LASANA and JIGSAWS datasets. Each evaluated method serves a targeted diagnostic purpose: end-to-end training to test whether supervised skill learning transfers directly, Adaptive Sharpness-Aware Minimization (ASAM) to probe whether flatter loss landscapes improve generalization, and augmentation-based self-supervised and contrastive learning to assess whether domain-invariant pretraining decouples skill from visual context. Transfer is evaluated in both directions using a disjoint-participant held-out test set for JIGSAWS. Results reveal an asymmetry: backbones pretrained on JIGSAWS achieve CCC values of 0.77 to 0.80 on LASANA, closely matching the end-to-end baseline, showing cross-rubric transfer is feasible when the target domain provides consistent supervision. Transfer to JIGSAWS fails across all methods, likely due to annotation inconsistencies. Control experiments with a Kinetics-pretrained backbone suggest task-specific heads carry the majority of the skill prediction burden, while the backbone need only provide adequate spatiotemporal features. These findings offer a new perspective on vision-based skill assessment: the central question of whether skill representations transfer across scoring systems has not been previously investigated. Results indicate the visual component is dominant but not solely responsible for skill prediction; further work is needed to conclusively disentangle transferable skill features from those bound to a specific visual domain.

Gender bias or other social biases in large language models (LLMs) are frequently evaluated with question answering or survey benchmarks where the LLM needs to give a response in a predefined answer format. It is well known in survey science that the answer format has a substantial impact on answers, just as LLMs are sensitive to the prompt wording. However, to our knowledge it has not been studied yet how changes in answer format impact the measurement of gender bias in LLMs and their alignment with human response distributions. We evaluate three instruction-tuned models on the BBQ benchmark and OpinionQA survey data across closed-ended, Likert-scaled and open-ended formats, comparing bias measurement and distributional alignment under otherwise identical conditions. We find that answer format does substantially alter measured outcomes, including reversals in order rankings. These differences arise because each format elicits distinct response behaviours, such as forced-choice selection, scale-based distributions and refusal in free-text generation. Our findings highlight the importance of treating answer format as a substantive component of LLM evaluation and motivate multi-format designs for more robust model assessment.

Background: Large Language Models (LLMs) are increasingly being applied to Software Engineering (SE) tasks, achieving high accuracy across problems such as clone detection, vulnerability prediction, and code summarization. However, their high computational demands and energy consumption raise sustainability concerns and hinder their use on consumer hardware and resource-constrained platforms. A common way to report the computational cost of an LLM in the literature and industry is to use the number of Floating Point Operations (FLOPs) required to perform a pass over the network. Aims: This paper investigates the implications of energy-aware knowledge distillation for SE, aiming to improve model efficiency while maintaining performance and to determine whether FLOPs is a reliable energy-aware metric. Method: We conduct a controlled experiment using Morph, a Many-Objective Optimization-based distillation methodology, to empirically examine whether FLOPs accurately reflect energy consumption in Clone Detection and Vulnerability Prediction tasks. We extend this methodology to include energy-surrogate models that directly estimate CPU and GPU energy consumption during optimization, and we apply Morph to generative tasks using CodeT5+ for code summarization. Results: Our results show that FLOPs is not always a reliable indicator of energy consumption, and better results can be achieved by using energy-surrogate models. Distilled student models can reduce inference energy consumption by up to 90\% and memory usage by 86\%, with only modest accuracy trade-offs. Conclusions: Energy-aware knowledge distillation when guided by direct energy surrogates rather than FLOPs can improve the energy consumption, sustainability, and deployability of LLMs for SE applications, enabling efficient models on consumer hardware.

Agentic porting of multiple open-source libraries into Rust, with human in the loop, has been performed for construction of modules within the Open-source Unified TRAnsient Multi-Phase Advanced Reactor simulation Kit (Outram Park). With this new methodology, verification and validation with human expertise, rather than code generation has become the bottleneck in developing reliable simulation codes. In this work, we present the porting of OpenFOAM libraries into the Outram-Foam Rust libraries, their preliminary verification and validation (V\&V) efforts, and their subsequent use in the development of open-source two-phase homogeneous-equilibrium (HEM) choked-flow solvers for the Thermo-hydraulic AI Multi-Phase INtegrated Emulator System (TAMPINES) libraries within Outram Park such as tampines-steam-tables. Preliminary V\&V efforts of Outram-Foam show that the cavity and Sod shock tube cases agree reasonably well with literature values. Moreover, the preliminary development of tampines-steam-tables shows good agreement with Moody's HEM charts. Thereafter, the 1D HEM solver, developed agentically, is presented and preliminarily validated against the Edwards blowdown case. Productivity increases were observed with the use of Claude Code, but domain expertise supplied by human experts remains critically required to ensure the generated code can solve the problem effectively. Further work remains to be done in V\&V, but the agentic coding methodology in this paper demonstrates great potential to speed up production and development of open-source libraries such as Outram Park.

Recent efforts toward fully automated AI scientists have demonstrated that language-model agents can generate hypotheses, execute experiments, and draft scientific manuscripts. However, during the early stages of research, when research problems are formulated, these AI scientists often rely heavily on proprietary frontier models. Their proposals are shaped by opaque parametric knowledge and by literature searches conditioned on the proposals themselves. Such knowledge is effectively a black box, and this dependence makes the evidential basis and validity of generated research problems difficult to audit and leaves the process vulnerable to model-specific hallucinations and biases. Furthermore, if proprietary research materials are transmitted to external APIs, the use of these models creates confidentiality, privacy, and data-governance concerns. We introduce the Structural Gap Hypothesis Agent (SGHA), a fully automated, corpus-first research-problem discovery system that runs entirely on a local LLM. SGHA structures a scientific literature corpus into evidence-linked paper objects and a typed evidence graph, detects unresolved structural patterns across papers, screens candidate gaps before formulation, and produces traceable research-problem families. In particular, it is able to output assumptions, objectives, success criteria, and remaining ambiguities. All LLM-based components of SGHA are executed using a locally served open-weight 9B language model, without requiring proprietary frontier-model APIs. We compare SGHA with the AI Scientist-v2 idea formulation module in five machine-learning domains. Our results suggest that explicit corpus structure and evidence-constrained reasoning can support promising, inspectable research-problem formulation without relying on frontier models during generation or verification.

User-facing tool agents must coordinate dialogue and tool use as user goals unfold over multiple turns. Yet interactive reinforcement learning typically reduces each rollout to a terminal reward, assigning the same credit to effective elicitation, errors, and later repair. The next user turn is more than context: it also provides noisy, temporally local evidence about the preceding user-to-user segment. We introduce \textbf{F}eedback-\textbf{A}ware \textbf{C}redit \textbf{A}ssignment (\textsc{FACA}), which aligns each reaction with that segment, derives a locally normalized reaction advantage, and adds it to verified terminal outcome advantage without an extra critic or rollout. Against an outcome-only Interactive GRPO control matched in simulator, visible dialogue, initialization, rollout, and optimization, \textsc{FACA} improves the nine-domain $τ$-family average across three independently trained runs by 5.91 and 10.22 percentage points at 8B and 14B, respectively. Gains concentrate in Telecom; at 8B, randomizing reaction polarity removes the Telecom gain. The same ordering holds zero-shot on Pare-Bench and Co-Gym. These results demonstrate that next-turn user reactions provide actionable local credit for improving multi-turn user-interacting agents.

Recent advances in LLM agents have made them increasingly capable of designing methods for complex AI tasks. This raises two central questions about agent-designed methods relative to human-designed methods: how well they perform, and how different their algorithmic designs are. To study these questions, this paper introduces an analysis that derives task-specific algorithmic design spaces from human-designed methods, maps both human- and agent-designed methods into these spaces, and quantifies their algorithmic differences at the module level. Widely used LLM agents are evaluated on a suite of representative, open-ended AI tasks spanning multiple modalities, and the methods they design are analyzed in terms of both task performance and algorithmic differences from human-designed methods. Experimental results show that current agents can occasionally match or surpass human state-of-the-art (SOTA) performance (10/72 configurations), but such success does not generalize reliably across tasks or agents. Moreover, 96.8% of agent-designed methods fall within human-derived algorithmic design spaces, largely recombining algorithmic choices found in human-designed methods, while nearly half exactly match an existing human algorithmic design. Taken together, these findings suggest that although current agents can occasionally match or surpass human SOTA performance, their algorithmic designs remain within human-derived algorithmic design spaces, reflecting the reuse and recombination of algorithmic choices.

Storyboards turn screenplays into visual shot plans for automated short drama production. Professional storyboarding relies on tacit directorial expertise and remains an industrial bottleneck. Large language models can automate this step, but methods for supplying directing knowledge face three challenges: (1) Knowledge acquisition: the craft remains implicit in exemplars or must be written manually. (2) Knowledge refinement: authored knowledge is not evaluated against execution outcomes, and opaque generation prevents feedback attribution to the knowledge behind each decision. (3) Knowledge injection: injecting all knowledge exceeds usable context, while manual selection for every narrative group does not scale. We present SAGE (Skill with Attribution-Guided Evolution), a deployed framework that learns, attributes, evolves, and routes directing knowledge from expert demonstrations. SAGE derives rules that are independent of episode content by contrasting each training screenplay with its expert storyboard. During generation, the model records each narrative group's adopted rules. Combining these records with localized feedback enables targeted updates to individual rules. Evolved rules form scenario packages with a routing index, so each group retrieves only a bounded set appropriate to its situation without expert intervention. On 18 test episodes across three genres, SAGE scored 77.8 on a rubric validated by experts, versus 77.1 for professional directors. Deployed for 14 days on Virtual Film Studio, SAGE produced 1,344 narrative group outputs; 87.2 percent were accepted without substantive edits, and the production team recorded over 83 percent less authoring time per episode. We release PROSE, the first public dataset pairing screenplays with storyboards by professional directors across 68 episodes: https://github.com/creDreams/PROSE.

Regularized sparse regression has been extensively studied in the offline setting, but online formulation remains relatively under-explored. This gap stems from four key challenges: (i) the infeasibility of dynamically updating the regularization parameter in every online round, (ii) managing storage and memory complexity, (iii) enabling real-time computation via closed-form updates rather than solving full optimization problems at each round, and (iv) achieving optimal statistical guarantees under realistic assumptions. In this paper, we propose an online generalized-sparsity-constrained regression framework, focusing on online cardinality-constrained linear regression and low-rank matrix sensing. Unlike online regularized regression, our constrained formulation eliminates the need for dynamic parameter tuning. We introduce an efficient online hard-thresholding algorithm that performs closed-form updates and requires storing only summary statistics, making it computationally, memory, and storage efficient. Despite the inherent nonconvexity and combinatorial nature of the formulation, our algorithm achieves global convergence at the optimal statistical rate under realistic assumptions, provided that the projection set is properly overparameterized. Numerical experiments demonstrate that our method consistently outperforms state-of-the-art alternatives.

This paper presents a pipeline for analyzing media bias and framing in online news. The pipeline groups articles into topics and events, adds named-entity and sentiment annotations, and compares news sources through people mentions, source-level tone, and event-level coverage patterns. We apply it to 8,358 Albanian news articles collected from GDELT and compare the resulting annotations with GDELT's automated annotations. The results show moderate agreement for sentiment and entity extraction, as well as additional person-entity pairs that can potentially support the bias analysis. We compare two annotation prompts and find that stricter sentiment-validation rules remove label-score inconsistencies but increase execution time and reduce annotation coverage. Based on these results, the simpler prompt is used for the rest of the analysis. We have provided sample analysis on source-level framing pro les, person-level tone differences across sources, and event-level gatekeeping and coverage indicators. These outputs show how the same news collection can be used to examine what sources cover, how they describe public figures, and where coverage is concentrated. The approach is particularly useful in settings where manually verified datasets or specialized language tools are limited.

Machine learning methods predict many real-world systems with remarkable accuracy, but they are typically treated as black boxes that offer no insight into which interactions drive the dynamics. Causal discovery methods reconstruct the interaction network from observational data, but without regard to whether the inferred structure supports prediction. Existing approaches combining both tasks rely on a single global hyperparameter, such as a causal threshold or a fixed neighborhood size, which cannot recover the structure of heterogeneous systems. Here we introduce causal local states (CLS), a framework that simultaneously infers an approximate Granger-causal interaction network and forecasts the system dynamics. For each node independently, we select the smallest set of neighbors that allows a predictive model to forecast the node near-optimally, and the resulting neighborhoods are then combined for a forecast of the full system. On three benchmarks of increasing difficulty, we achieve reconstruction of the underlying networks with high fidelity and forecasts on par with a model that is supplied with the true network, providing a step toward explainable and scalable forecasting of complex systems.

Learning restricted Boltzmann machines (RBMs) is computationally challenging because it requires expectations whose exact evaluation is generally intractable. The expectations are typically evaluated using a sampling approximation based on blocked Gibbs sampling (BGS), which is a local Markov chain Monte Carlo transition kernel. However, the locality of BGS can lead to poor sampling quality when the RBM has high energy barriers, thereby degrading learning performance. Deep tempering (DT), which performs parallel tempering over a sequence of learnable RBMs including the training RBM, alleviates this locality issue. However, DT algorithmically requires multiple steps to move through the RBM sequence to achieve a nonlocal transition. In this paper, we propose a transition kernel defined over the RBM sequence used in DT. The proposed kernel has a round-trip structure over the sequence, enabling nonlocal moves within a single transition while leaving the RBM sequence invariant. Numerical experiments show that the proposed kernel performs nonlocal transitions more frequently and achieves higher sampling quality with fewer transitions than BGS and DT. We further verify that learning based on the proposed kernel is more stable and mitigates the training failures observed with BGS- and DT-based learning.

Most large language model services use stateless defenses, which judge only the current request, to refuse harmful tasks. Decomposition attacks exploit this limitation by splitting a harmful task into individually permissible requests and combining their answers. Defending against them therefore requires a stateful monitor that considers requests together. If it can group all requests for one attacker task, it can stop the attack. However, attackers can use unlinkable identities and combine answers elsewhere, leaving no reliable grouping signal. We ask whether decomposition attacks can still be stopped under this setting. For a fixed attack strategy without retries, we prove that the achievable security and utility tradeoff depends entirely on how benign requests for the same capabilities are grouped. Persistent, recognizable groups permit a useful defense; fresh, indistinguishable groups do not. When attackers can retry and learn from Allow/Block decisions, this useful operating point disappears: the feedback reveals what passes but not whether a block was correct. Experiments on 91 executable tasks and 11,393 capability-matched benign requests support these results. Under a 1% denial cap for these requests and a 0.5% cap for unrelated background traffic, all ten tested policies, including one privileged policy with an exact request-to-operation map, either fail to stop attacks or exceed the budget. On defense-unseen task families, attack success is at least 99% after one attempt and 100% after two. Effective defenses therefore require additional evidence or mechanisms tied to grouping, such as reliable identity linkage, costs for fresh identities, or control over answer use.

Knowledge Graph Reasoning (KGR) aims to discover latent facts by leveraging the structural evidence available in KGs, posing a challenge to the structural semantic understanding capability of KGR models. Recent studies have demonstrated that Large Language Models (LLMs) can achieve remarkable progress on KGR tasks via flexible in-context learning. However, the inherent representation inconsistency between KG structural context and LLM parametric knowledge remains inadequately addressed. This limitation prevents LLMs from effectively perceiving reasoning evidence that aligns with KG constraints, which undermines both the effectiveness and faithfulness of reasoning. We refer to this problem as reasoning evidence perception drift of LLMs over KGs. To address this problem, we propose a Structure-Internalized Rule Language Model (SIRLM), which centers on structural rule generation to couple the parametric learning of structural knowledge with the faithfulness evaluation of reasoning logic, enabling LLMs to anchor tightly to KG-grounded evidence. Specifically, we first design a Structure-Internalized Rule Generator (SIRG), which incorporates an in-context learning block augmented with a structural relation memory to coordinate structural and parametric knowledge. Furthermore, we equip SIRG with a KG tokenizer based on structural invariance learning and a neuro-symbolic reasoner based on rule-constrained message propagation. These components provide SIRG with learnable structural representations and faithful rule-execution feedback, respectively. Our SIRLM can be seamlessly integrated into standard LLM training paradigms, such as SFT and GRPO. Extensive experiments against 17 state-of-the-art KGR methods on 36 datasets demonstrate the significant superiority of SIRLM.

Although Graph Neural Networks (GNNs) have made significant advances in spatio-temporal traffic forecasting, their performance is limited when they rely solely on sensor proximity or road-network topology. This paper presents a spatio-temporal prediction framework, developed to incorporate knowledge in various forms. This framework aims to improve sensor-level, contextual understanding of the environment. A general-purpose knowledge graph (e.g., Wikidata) is used to create semantic subgraphs around traffic sensors and generate knowledge graph embeddings that capture meaningful relationships, such as nearby points of interest, administrative hierarchies, and the functional roles of locations. These embeddings are then fused with conventional traffic sensor graphs to provide additional adjacency matrices informed by semantics. This allows GNNs to learn the semantic context beyond physical connectivity. This study differs from previous research in two key ways. Firstly, rather than proposing a novel GNN architecture, it demonstrates the general impact of external knowledge on prediction accuracy. Secondly, experiments with well-established traffic forecasting approaches show that external knowledge provides additional information that street network data alone cannot convey. The results show that integrating data from general-purpose knowledge graphs and sensor networks through data fusion can enhance the prediction accuracy of traffic forecasting models, and offers a potential pathway toward improved interpretability.

We study Gaussian regression over the explicit vector-valued Parhi--Nowak deep-RBV^2 architecture with depth L, width w, layer-sum variation budget A, and output bound B. For this O(L w^2)-parameterized architecture, the known lower and upper bounds differ by one factor of depth. We construct a local packing showing that the quadratic depth dependence is intrinsic under an explicit sample-size-dependent radius condition. The packing has log-cardinality Omega(L^2 w^2 log w); its codewords lie in an O(lambda) L^2 ball and are pairwise Omega(lambda)-separated. The main ingredients are a bias-corrected bounded-coefficient approximation theorem and balanced amplification: multiplying a depth-D ReLU network by q can be implemented using one constant channel so that every coefficient grows by only q^(1/D). Translation to vector-valued RBV^2 blocks then has layer-sum cost O(D w^2 q^(1/D)). Gaussian Fano yields a radius-explicit lower bound governed by the output, testing, and representation scales. Under A=B=R, sigma proportional to R, and the stated radius condition, this gives minimax risk at least of order L^2 w^2 log(w) R^2/n. A pseudodimension-based finite-net upper bound gives O-tilde(L^2 w^2 R^2/n) for unbounded Gaussian responses. Thus the minimax risk has quadratic polynomial dependence on depth, up to logarithmic factors, and exhibits a transition to representation-limited behavior at smaller radius.

LLM agents have been widely adopted to operate mission-critical infrastructure (MCI). These agents normally rely on a harness that determines what information they can access, which tools they can use, and what actions they can take. Existing systems often expose the same comprehensive harness to every task, which may not be necessary and cause resource wastes. In this paper, we focus on the identification of optimal harness configurations, and view it as a resource-matching problem between what each task requires and what the harness provides. To measure this match, we classify MCI tasks based on the mathematical representation of the underlying system and rank harness configurations by the amount and type of information they provide. We then construct task-to-harness mappings from two sources: mining research literature and measuring controlled agent execution. Leveraging the measured mapping, we propose a new harness provisioning algorithm: map-guided escalation. It begins with a task-specific harness and expands to full provision only after a failed self-check. We evaluate our method in two representative MCI tasks: in liquid cooling, it improves the agent accuracy from 0.652 under full provision to 0.715 and achieves accuracy comparable to Reflexion with 48% fewer tokens; In power grids, full provision remains accuracy-optimal, while map-based provisioning offers lower-cost alternatives. These findings show that harness provisioning follows a domain-dependent accuracy-cost Pareto frontier rather than a universal optimum.

We introduce Semantic Task Completion Video Generation, an outcome-oriented video generation task. Under this formulation, success requires both achievement of the intended outcome and semantic grounding. Semantic grounding characterizes the correspondence between the reference image and the generated outcome in terms of high-level semantics relevant to the task. Evaluation focuses on the generated outcome and requires neither the presentation of a complete sequence of intermediate task steps nor conventional appearance consistency with the reference image. To support systematic evaluation, we construct SemComp-Data, an evaluation dataset covering six domains. Each instance comprises a reference image, a detailed instruction, a brief instruction, and an outcome-centric video clip. A scalable four-stage curation pipeline converts raw videos into standardized SemComp-Data instances. We further introduce SemComp-Bench, an evaluation protocol that uses a vision-language model (VLM) to answer structured binary questions. SemComp-Bench reports the OA Score and the GR Score for Outcome Achievement and Generation Reliability, respectively. Experiments on representative video generation models show that achieving intended outcomes while maintaining task-relevant semantic grounding in reference images remains challenging.

GRPO is increasingly used for reinforcement learning of vision-language-action (VLA) policies because, unlike PPO, it does not require training a critic. This simplification comes with a sampling cost: group-relative advantages require multiple rollouts from each scene. Under binary success rewards, groups whose rollouts all succeed or all fail have zero advantage and are discarded by dynamic sampling. These groups are especially common early in training, when most rollouts fail, wasting much of the expensive robotic rollout budget. We introduce Prism-GRPO, which augments binary outcome reward with a weighted trajectory-level execution-quality score. By splitting same-outcome groups into a quality spectrum, Prism-GRPO recovers training signal while ensuring that every success still outranks every failure. Quality scores can be derived from simulator contacts, executed actions, or visual observations, avoiding task-specific progress rewards. We prove that Prism-GRPO never increases the probability that a sampled group is discarded for having zero advantages, and derive a gradient-alignment condition under which its combined update remains a local ascent direction for task success. Across four RoboTwin tasks spanning different horizons and coordination patterns, Prism-GRPO improves success and quality at matched rollout budgets and reaches target success rates with up to 56% fewer rollouts. It also suppresses a reward-hacking shortcut, with the cleaner behavior transferring under direct deployment to a real robot. Through ablations, we show consistent gains across contact-, smoothness-, and VLM-derived quality signals.

Group Relative Policy Optimization (GRPO) has become a widely used approach for post-training Large Language Models (LLMs) for reasoning. In GRPO, the group gradients induced by different queries within the same mini-batch are directly averaged to form the policy update. However, these group gradients can point in conflicting directions. Our empirical analysis suggests that group-gradient conflicts tend to be associated with less effective policy updates, motivating the need for a reliable aggregated update direction under such conflicts. Standard GRPO aggregation treats the realized group gradients as deterministic contributions and does not account for differences in their reliability during aggregation. To address this issue, we propose Gradient Uncertainty-Aware Policy Optimization (GUPO), which models each group gradient as a random variable under a Bayesian formulation and estimates its probability distribution. GUPO then derives gradient uncertainty using a Dirichlet-based formulation and uses it to calibrate the contribution of each group gradient during aggregation. Extensive experiments on multiple benchmarks demonstrate the effectiveness of GUPO.

Modern software systems evolve through frequent commits that implement bug fixes, features, and security patches. Although code review and testing are widely used to check these changes, they often provide limited assurance for memory-safety issues. Code reviewers may miss subtle boundary, lifetime, or initialization errors, while existing tests may not exercise the specific paths affected by a commit. Fuzzing is effective at exposing such bugs, but applying it to every commit remains impractical because whole-program fuzzing is expensive, requires suitable harnesses, and may still fail to reach the code changed by a commit. In this paper, we introduce COMMITGUARD, a commit-aware differential slice-based fuzzing approach for verifying code changes. The key insight behind COMMITGUARD is that the pre-commit version of a modified function can serve as a behavioral baseline for interpreting bugs found after the commit. For each target commit, COMMITGUARD identifies modified functions, extracts compilable code slices from both the pre-commit and post-commit versions, and fuzzes the paired slices independently. It then compares sanitizer reports across the two versions and reports bugs that emerge only in the post-commit version as candidate commit-induced bugs. We evaluate COMMITGUARD on 300 commits from openSSL, libpcap and leptonica. Slice fuzzing initially produces 518 sanitizer reports across these commits. By comparing pre-commit and post-commit slices, COMMITGUARD narrows this large output to 7 candidate commit-induced bug reports that require manual triage. Manual validation confirms 5 of these reports as real bugs that were fixed by developers of the examined projects after we reported them, while only 2 reports were classified as false positives. COMMITGUARD analyzes a commit in 32.4 minutes on average and achieves 75.36% average coverage of modified functions.

Text which has been translated from another language tends to carry with it evidence of translation$\unicode{x2014}$hence, it is often referred to as $\textit{translationese}$. Multilingual large language models (MLLMs) generate text in a variety of languages. However, it is still unclear if MLLMs' generations resemble internal translation (from English or, potentially, other languages) and, thus, result in translationese. Here, we ask the following research questions: (1) Does text generated by MLLMs resemble translationese? (2) How does translationese produced by MLLMs differ from translationese produced through direct translation? We leverage established indicators of translated text to evaluate text generated by state-of-the-art MLLMs in five languages, comparing to both non-translated and human-written baselines in order to isolate translationese from other kinds of interference. Through the use of high-accuracy classification models, analyses of variance on individual linguistic features, and the collection of human annotations in a subset of two languages (German and Spanish), we assess the translationese content of MLLM generations and examine the key features that distinguish MLLM-generated text from typical translation-related interference.

Modern software systems are increasingly complex, and static analysis tools are commonly used to identify potentially vulnerable code by issuing warnings. However, these warnings often require manual inspection to confirm whether the reported issues are real, making the process time-consuming and error-prone. Directed fuzzing has emerged as a powerful automated technique to validate the warnings. However, applying it to the entire project in response to each warning is computationally infeasible, often requiring days of execution to achieve only incremental improvements in code coverage. We present SNIPTEST, an execution-based warning triage framework that generates and fuzzes compiled code slices centered around static-analysis warnings. Rather than proving exploitability in the full program, SNIPTEST provides evidence about how a warning behaves under progressively expanded sliced execution contexts. It employs a layer-by-layer slicing strategy, incrementally expanding context around the target location to validate potential vulnerabilities with increasing precision. We evaluate SNIPTEST on a benchmark of 97 true vulnerabilities and 97 false alarms across three real-world projects. SNIPTEST produces Possible True Positive evidence for 53 of 97 confirmed vulnerabilities (54.6%) by triggering the corresponding bug oracle consistently across all three analyzed slice levels, while the remaining cases are unreachable. Particularly, in 40.2% of these cases, it exploits the vulnerability along the observed execution path, matching the top three stack frames. On the 97 confirmed false alarms, SNIPTEST produces Possible False Positive evidence for 54 cases (55.6%) by reaching the warning without triggering the bug oracle, but misclassifies 28 cases (28.8%),and the remaining cases are unreached. Finally, we demonstrate the practical relevance of SNIPTEST by identifying CVE-2025-11964.

Reinforcement learning for coding agents increasingly relies on long-running agent harnesses to manage tool integration, repository contexts, and execution feedback. However, the native execution environments of these harnesses are inherently misaligned with policy-gradient training: environmental crashes and reward hacking corrupt outcome signals, while train-inference discrepancies decouple rollout behavior from policy updates. To address this, we present LEGO-RL, a framework that bridges native coding-agent harnesses with scalable policy-gradient optimization without modifying their internal control flow. LEGO-RL is built upon three pillars: (1) faithful optimization via in-process LLM proxying that captures raw generation streams for token-level alignment and robust trainer-side log-probability recomputation, even under harness-side compaction or re-serialization; (2) reliable execution via scalable sandbox orchestration featuring image caching and stage-wise defenses to mitigate reward hacking; and (3) observable training through an integrated plugin that automates validation and monitoring, paired with a Live UI for granular trajectory diagnostics. We evaluate LEGO-RL by training the sparse MoE model Qwen3.5-35B-A3B with GSPO across three native coding-agent harnesses. LEGO-RL improves Qwen3.5-35B-A3B across OpenHands SDK (64.0% to 70.4%), Claude Code (62.4% to 68.2%), and OpenCode (57.2% to 66.6%) on SWE-bench Verified, while maintaining a rollout-training probability correlation above 0.99.

Biomolecular design underpins applications from molecular recognition to therapeutics and synthetic biology, yet de novo interaction design remains challenging-especially for DNA/RNA, underexplored non-protein modalities with scarce, heterogeneous complex data and sharper geometric and chemical constraints. We introduce MCTH (Monte Carlo Tree Hallucination), an inference-only framework that casts all-atom sequence-structure co-design as uncertainty-aware planning over hallucinated states from pretrained folding and inverse-folding models, with optional biophysical control within the same decision loop. MCTH treats these models as frozen black-box operators and uses Monte Carlo Tree Search to allocate a fixed inference budget across competing design trajectories, incorporating model confidence and uncertainty, as well as cross-expert consensus/disagreement when multiple predictors are available. Across protein-RNA, protein-DNA, protein-protein, and protein-ligand design, matched-budget experiments show that adaptive search improves over simpler sampling and cycling strategies, while held-out AlphaFold3 and Chai-1 evaluations demonstrate transfer beyond the search-time oracle. MCTH provides a shared planning layer across modalities while allowing task-specific folding, inverse-folding, and biophysical modules, requiring no fine-tuning or backpropagation through component models.

We introduce PTXBench, a benchmark for evaluating and adapting large language models (LLMs) to use architecture-specific PTX for GPU kernel optimization. PTXBench measures functional correctness, whether selected target instructions execute at runtime, and speedup over frontier libraries across GEMM and attention workloads on H100 and B200 GPUs. Our evaluation shows that architecture-specific PTX capability remains uneven: success rates fall substantially on complex attention backward workloads, and executing the target instructions does not necessarily translate into competitive performance. No evaluated model consistently matches frontier libraries across the suite. We further adapt Qwen3.6-27B using supervised fine-tuning. Repair-conditioned training improves several tasks, but generalization remains uneven; data coverage, balance, and the quality of the reasoning teacher matter in addition to dataset size. PTXBench provides an auditable testbed for measuring and improving LLMs' ability to exploit evolving GPU architectures.

Motivated by a conjecture of Vaikuntanathan and Zamir, we study the pseudo-mixing of Kac's walk on $\mathrm{SO}(n)$: whether short trajectories are indistinguishable from Haar measure by low-complexity tests. We prove that the first $k$ columns mix in Wasserstein distance in $O(n(k+\log n)\log n)$ steps for fixed accuracy, resolving a conjecture of Oliveira. Combining this with a representation-theoretic variance bound, we show that if $T=ω(nk(k+\log n)\log n)$, then every degree-$k$ polynomial normalized to have unit Haar variance has expectation under the $T$-step law within $o(1)$ of its Haar expectation. As an application, we show that this pseudo-mixing estimate can be used to prove the effectiveness of a fast Johnson--Lindenstrauss transform with the usual target dimension.

Sample efficiency is a central challenge in reinforcement learning (RL), particularly in image-based domains where agents must learn from high-dimensional visual inputs. Traditional sampling often relies on random or suboptimal experience selection, leading to redundant updates and slow learning. Improving efficiency requires mechanisms that prioritize informative experiences while also encouraging effective exploration. Prioritized Experience Replay (PER) addresses part of this challenge by reusing high-value transitions, while intrinsic rewards promote the exploration of novel or uncertain states. However, their integration has not been extensively studied. This paper introduces Novelty and Surprise Prioritized Experience Replay (NSPER), which uses novelty to capture underrepresented states and surprise to expose gaps in the agent's understanding of the environment. We further extend this with NSPER+R, integrating these signals as intrinsic rewards to jointly improve replay quality and exploration. Experiments on DeepMind Control Suite tasks show that NSPER and NSPER+R improve training efficiency and convergence speed compared to existing methods in image-based RL.

Generative models promise a route to explainable clinical AI: rather than probe a classifier, model the distributions of healthy and diseased patients and read explanations off the geometry between them. We build such a system - an optimal-transport rectified flow trained between two clinical distributions - and use it to ask a pointed question the field too rarely tests: do the resulting explanation heatmaps actually localize disease? On tabular tumour biomarkers (Breast Cancer Wisconsin) a single flow yields per-patient counterfactuals, an unsupervised malignancy score (AUROC 0.91; 0.93 +/- 0.01 across five seeds), and a label-free attribution that agrees with a supervised classifier (r ~ 0.5) - a compact, honest interpretability engine, though it never out-predicts logistic regression. Moving to chest X-rays, we show the transport heatmap is a population-level signal, not a localiser; a reconstruction-based, identity-preserving variant does localize synthetic lesions (pointing game 0.52), yet on real RSNA radiologist boxes it collapses to chance while only supervised Grad-CAM stays above it. The central result is a synthetic-to-real gap: label-free heatmaps that look compelling on planted lesions are not evidence of real localisation. We contribute a reusable optimal-transport recipe for generative explanations and a controlled benchmark for stress-testing whether they localize.

Merging finetuned models combines specialized capabilities without joint training or access to the original data. Most methods operate by linear arithmetic in Euclidean weight space, which cannot carry the geometry of the update. Orthogonal Model Merging (OrthoMerge) uses a single orthogonal transform for each weight matrix, but such a transform cannot change singular values. We propose CORAM, which partitions each target matrix into row slices, represents every expert slice by its singular value decomposition in the corresponding base-model SVD frame, and merges the task-specific factors on their corresponding manifolds. Because manifold averaging contracts the merged update, CORAM applies an amplification coefficient $λ=κ\hat{c}$. The scale c_hat is estimated from the expert and merged update norms and is approximately $\sqrt{N}$ for $N$ experts with comparable update magnitudes. The restoration strength kappa is selected from the dispersion of expert updates without evaluating candidate merged models. This rule remains within 0.72 points of the best swept value on all evaluated suites. CORAM also includes spread slicing to distribute highly updated rows across slices and a residual pathway for non-target layers. Across four suites covering three model families, 3B to 9B scales, and language and vision-language experts, CORAM improves over OrthoMerge by 0.25 to 1.35 points and matches or exceeds the strongest weight-space baselines.

Reliable jailbreak evaluation is essential for assessing LLM safety, but most existing studies rely solely on attack success rate (ASR) without accounting for its dependence on attack budgets, resulting in unfair comparisons across methods. Existing compute-aware evaluations reduce heterogeneous resources into FLOPs, which is difficult to estimate for black-box models and fails to capture resource-specific constraints. To provide a comparable evaluation basis, we introduce Fair-ASR, an evaluation protocol for black-box jailbreak attacks under shared target-call budgets B, using target calls as a directly observable and method-agnostic comparison axis while tracking attacker calls separately for efficiency analysis. We re-evaluate 11 representative attacks under the Fair-ASR protocol and find that attack rankings change substantially across target-call budgets, simple stochastic perturbations and hand-crafted templates remain highly competitive under equal target access, and no evaluated LLM-driven method is efficient in both target and attacker calls. Motivated by this efficiency gap, we introduce ReCode, a compositional budget-efficient attack that combines desensitization rewriting with two effective low-cost primitives identified by Fair-ASR. Under a budget of 20 target calls, ReCode achieves 85% ASR on GPT-5 while requiring only 7.19 attacker calls per request on average, showing strong efficiency in both target and attacker calls.

Most automated essay scoring (AES) systems output a single holistic score without interpretable evidence and rely on closed APIs that introduce data privacy and cost barriers. We present ArguLens, an opensource, locally deployable system that decomposes AES into three decoupled components: a discourse-move classifier (Qwen2.5-7B-Instruct fine-tuned with LoRA on PERSUADE 2.0), a grade-independent LightGBM scorer over 31 linguistic and discourse features, and a label-aware feedback generator served through vLLM with a Qwen2.5-14BInstruct backbone. A Gradio web UI exposes pluggable inference backends and supports single-essay and batch scoring with downloadable per-essay breakdowns. On an essaydisjoint PERSUADE 2.0 test split, the logitprobe classifier achieves 82.6% accuracy and 0.727 macro-F1; under prompt-grouped 5-fold cross-validation the scorer reaches a mean QWK of 0.813 under an oracle discoursefeature protocol, and an ablation shows that adding gold discourse annotations yields an increment of +0.055 QWK over the lexical+syntactic configuration (paired t-test, p = 0.010). This is a component-level diagnostic rather than an end-to-end classifier-to-scorer result. The feedback generator ships with a structured evaluation protocol; its human-rater study is left to future work. The system is released under Apache 2.0 at https://github.com/wwrwbs/AI_AWE.

Distributed Denial-of-Service (DDoS) attacks threaten network availability, requiring a cognitive detection process that senses traffic, infers intent, and supports an adaptive response under severe class imbalance and non-stationary conditions. This paper proposes a Graph-based Generative Adversarial Network (GraphGAN) that serves as the cognitive detection engine for this task. GraphGAN captures the relational structure among traffic flows while addressing imbalance through adversarial generation of synthetic samples. Sequential flows are converted into $k$-nearest neighbor graphs using sliding windows to preserve feature-similarity and temporal dependencies among flows. The generator learns the distribution of DDoS attacks to synthesize realistic minority samples, while a Graph Convolutional Network (GCN)-based discriminator distinguishes real from synthetic graph data. A separate GCN classifier, trained on the balanced dataset, performs the final detection decision. Evaluations on four benchmark datasets show that GraphGAN achieves superior accuracy, precision, and recall compared to state-of-the-art approaches, particularly in data-scarce scenarios. By integrating temporal graph construction, adversarial augmentation, and GCN classification, GraphGAN effectively models coordinated attack behaviors and mitigates class imbalance, providing a robust and topology-aware solution for intrusion detection in data-constrained environments.

Repetition is a fundamental mechanism in human learning, where revisiting successful experiences strengthens memory, consolidates skills, and improves future performance. Motivated by this biological principle, we introduce Instant Episode Repetition (IER), a simple and novel mechanism that improves sample efficiency by immediately repeating action sequences from successful episodes during environment interaction. Unlike conventional approaches such as Experience Replay and Self-Imitation Learning (SIL), which passively reuse past experience during training updates, IER directly influences the data collection process. Upon identifying a high-reward episode, the agent repeats its action sequence for a fixed number of subsequent episodes, reinforcing valuable behaviors through renewed interaction with the environment. We integrate IER into state-of-the-art SAC and TD3 algorithms and evaluate its effectiveness on continuous-control benchmarks, including MuJoCo, the DeepMind Control Suite, and a real-world dynamic object translation task with a robotic manipulator. Experimental results demonstrate that this simple mechanism improves learning performance over standard and self-imitation-based baselines.

Neural network surrogates are an emerging alternative to traditional electromagnetic wave simulators like finite-difference time-domain (FDTD); their goal is to replace rigorous physical simulations with pre-trained neural networks that solve wave-scattering forward and inverse problems orders of magnitude faster. However, nonrecurrent, single-step surrogates have scaled only to a few tens of simulation variables. Here, we show that this barrier can be overcome by dynamically generating salient training examples during training, rather than randomly sampling the large space of possible examples. We introduce an algorithm that runs in parallel with surrogate training, using gradient ascent to search refractive-index and source configurations for cases where the surrogate disagrees with a full-wave ground-truth simulator. We also use source and ground-truth normalization with an evolving replay dataset to stabilize and accelerate learning. Using this approach, we train a fast, single-step surrogate for two-dimensional wave scattering with up to 41,772 controllable variables, including dense, freely configurable grids of refractive indices and complex-valued sources. The resulting neural surrogate is robustly accurate across diverse structured and unstructured examples and generalizes inductively to larger domains, reaching over 3 million controllable variables without retraining, a $73.8\times$ increase. We demonstrate the surrogate on large-scale forward simulations and inverse design of freeform beam splitters and gradient-index (GRIN) lenses up to 98 wavelengths wide, showing comparable or better performance than FDTD-based designs, with speedups from $1.29\times$ to $26.5\times$. These results demonstrate a practical path toward fast, robustly accurate, inductively scalable neural simulators for photonic inverse design and other wave-scattering inverse problems.

Data-driven algorithm design frames hyperparameter tuning as a statistical learning problem, but establishing generalization guarantees remains challenging due to the implicit, non-smooth dependence of model performance on hyperparameters. Existing multi-dimensional bounds under piecewise-polynomial assumptions remain theoretically loose and lack comprehensive lower bounds. We resolve this by establishing tight pseudo-dimension bounds for multi-dimensional data-driven tuning. First, we refine the learning-theoretic upper bound using real algebraic geometry; by analyzing invariant connected sign cells during block elimination rather than isolated sign vectors, we avoid topological over-counting to derive strictly sharper sample complexities. Second, we present a multi-regime lower-bound framework that disentangles combinatorial and algebraic capacities. By constructing shattered problem instances across distinct regimes, we prove our upper bounds are tightly saturated. Finally, we extend our topological framework to accommodate general bi-level validation-loss tuning and broader semi-algebraic applications.

Forecasting cryptocurrency prices remains a formidable challenge due to inherent non-stationarity, abrupt regime shifts, and multi-scale stochastic dependencies. Conventional deep learning models often struggle to capture complex underlying dynamics, frequently resulting in persistent phase-lagged predictions. To address these limitations, we propose MoFE, a novel deep learning framework that integrates Fourier Neural Operators (FNOs) within a Mixture-of-Experts (MoE) architecture. Rooted in the theoretical framework of stochastic differential equations, MoFE conceptualizes cryptocurrency volatility as a superposition of multi-frequency components, which includes user network based fundamental growth, mining costs and halving mechanism caused seasonal volatility, and market sentiment-induced chaos. Specifically, specialized adaptive FNO (AFNO) and Convolution dual-domain experts learn continuous function-to-function mappings to encapsulate global spectral trends, cyclical adjustments and microstructures, while a dynamic gating based MoE mechanism enables adaptive strategy switching across diverse market regimes. Extensive experiments on Bitcoin datasets spanning January 2020 to December 2025 demonstrate that MoFE achieves state-of-the-art (SOTA) performance in both T+1 and T+5 forecasting horizons. Notably, the model effectively mitigates the phase-lag effect, delivering superior Directional Accuracy (DA) and Information Coefficient (IC). In high-fidelity simulated trading environments, these predictive gains transfer into significant excess returns and robust risk-adjusted performance, characterized by a high Sharpe ratio.

AI planning is concerned with finding a sequence of actions that achieves a specified goal. It relies on explicit models of the world, commonly represented in the Planning Domain Definition Language (PDDL). An active line of research investigates how errors in such models can be detected and repaired. For example, users may provide positive test plans that are solutions, and negative test plans that fail during execution. Automated repair methods then modify the PDDL model to satisfy these constraints. In this paper, we evaluate the ability of recent open-weight large language models to perform this repair task using an LLM-only approach. Our experiments show that the symbolic baseline achieves an $F_1$ score of $.49$, while the best-performing LLM reaches $.87$ with high reasoning effort, an absolute improvement of $.38$. However, that setting has a mean test pass rate of only $.82$, falling to $.06$ on the Thoughtful domain; even the best setting that includes the test traces reaches only $.92$. Thus, current open-weight models cannot guarantee satisfaction of the test constraints required for reliable automated model repair.

Long-context prefill in large language models (LLMs) incurs substantial computation and memory traffic because dense self-attention computes quadratic query-key scores. Existing methods either use a uniform low-precision path or select token interactions, leaving spatial precision routing over hardware-aligned score tiles outside fused dense attention. We introduce TileMix, a tile-centric precision-routing kernel that makes numerical precision an executable spatial decision over score-tile groups within fused dense attention. TileMix partitions the attention matrix into hardware-aligned score tiles, packs routing decisions into compact bitmasks, and dispatches each tile group through FP16 or INT8 score computation while both paths update a shared online-softmax state. Scalable precision grouping lets each routing bit govern multiple adjacent key tiles, preserving hardware-aligned compute tiles and compact metadata at long contexts. By routing all legal tile groups, TileMix preserves dense token connectivity, requires no training, and supports grouped-query attention, variable-length batches, and INT8 key/value caches. Across LongEval, LV-Eval, and A100 prefill benchmarks on LLaMA, Qwen, and Vicuna, TileMix recovers long-context quality lost under uniform INT8 and improves prefill throughput over FP16, yielding a controllable accuracy-efficiency frontier across model families. The implementation is available at https://github.com/HanzhiZhang-Ulrica/TileMix.

Modern probabilistic time-series forecasters often express uncertainty through forecast samples. While typically converted into nominal prediction regions using empirical quantiles, these model-implied sets lack formal coverage guarantees and frequently deviate from nominal targets under distribution shift. Existing multivariate conformal methods can calibrate these regions online, but they typically estimate geometry from historical residuals using fixed or accumulating look-back windows. This reliance on the past limits their ability to exploit the instantaneous dependence structure of current predictions and leaves them vulnerable to stale-regime contamination. To address this, we propose SPACE, a conformal wrapper for sample-generating multivariate forecasters. SPACE constructs ellipsoidal joint prediction regions by estimating time-local covariance geometry directly from the current forecast sample cloud, calibrating the region's radius via a dynamic backward window-selection scheme. Across diverse multivariate datasets, probabilistic forecasters, and conformal baselines, SPACE consistently brings realized joint and rolling coverage closer to the nominal target, achieving superior coverage-efficiency tradeoffs relative to competing wrappers.

Large language models for medical consultation are often evaluated after a clinical problem has already been made clear, although real consultations may begin with a vague, minimized, or misframed concern. We evaluated three API models across four physician-authored, multi-turn vignettes under baseline and entry-to-care instruction conditions, yielding 24 fixed-script transcripts; two cases also used adaptive standardized-patient simulation, yielding 12 transcripts. Self-care or home-management advice before any patient answer appeared in 9 of 12 baseline case-model cells and 0 of 12 instruction cells, while structured handoff summaries appeared in 0 of 12 and 10 of 12 cells, respectively. The instruction changed sequencing and documentation, although it did not reliably ensure elicitation of decisive facts. The preformulation gap should therefore be evaluated directly through observable first-contact behavior rather than inferred from diagnostic accuracy or final-answer quality.

Tokenization is a fundamental component of language modeling pipelines. Despite its importance, it is often fixed, even though it significantly impacts model performance across languages. In this work, we analyze what tokens are learned when tokenization is jointly optimized with language modeling. We compare tokenizer-free approaches such as SSLMs and H-Nets with fixed tokenizers across 18 typologically and script-diverse languages. Our results show that joint optimization fundamentally alters token structure. SSLMs recover morphologically aligned and contextually efficient tokens, whereas H-Nets prioritize byte-level efficiency, producing longer tokens with very low overlap with standard subword vocabularies. We further show that tokenization behavior varies across language typologies. Agglutinative languages exhibit more dynamic segmentation patterns while learning. Through downstream evaluation, with pretrained-then-finetuned BERT models, we find that SSLM-based pretokenization consistently reduces language modeling perplexity and achieves competitive downstream performance despite distinct vocabularies. Overall, tokenizer-free approaches optimize for contextual and computational efficiency rather than strict morphological structure, resulting in fundamentally different yet effective vocabularies for downstream NLP.

Robotic manipulation policies trained via imitation learning, such as Action Chunking with Transformers (ACT), can achieve strong performance under ideal conditions but often remain sensitive to small execution errors and distribution shifts. Correcting these failures typically requires dataset aggregation and full-policy retraining, which is computationally expensive and unsuitable for real-time deployment. In this work, we propose Online Residual Policy Adaptation (ORPA), a framework that enables immediate, feedback-driven correction of robot actions without modifying the underlying policy parameters. ORPA augments a pretrained control policy with a lightweight, feedback-conditioned module that predicts residual adjustments directly in joint space, allowing the system to adapt its behavior at runtime. We evaluate ORPA on a set of precision-sensitive manipulation tasks using the ALOHA platform, demonstrating improvements in success rate and recovery from small perturbations compared to baseline control policies and rule-based inverse kinematics corrections.

Browser agents perform well on short, clean demonstrations, but real deployment is fundamentally different: agents must sustain dozens of decisions on live websites while recovering from mistakes and navigating complex UIs. We argue that closing this gap requires alignment at every level of the pipeline, including execution, supervision, optimization, and evaluation, rather than scale alone. We present Wuying-Browser-Agent, a unified framework that addresses each of these levels. A structured browser harness provides stable execution primitives and decision-oriented context management. Reflection and UI-specialized Curriculum SFT (RUIC-SFT) explicitly trains on recovery trajectories and complex-UI interactions. Divergence-Aware Online GRPO (DAO-GRPO) improves long-horizon credit assignment through potential-based reward shaping and divergence-aware step weighting. Finally, we introduce BrowserBench, a bilingual real-web benchmark of 350 tasks averaging 37.9 steps, because most existing benchmarks are too short to expose long-horizon failure modes. Wuying-Browser-Agent-27B achieves 80.6\% on WebVoyager, 66.7\% on Online-Mind2Web, and 65.1\% on BrowserBench, establishing a new open-source state of the art on browser-use benchmarks. The same pipeline also transfers beyond browser use, demonstrating strong general agentic ability and reaching an average score of 73.8 on Tau2-Bench, Claw-Eval, and BFCL-v4.

Reinforcement Learning (RL) has been promising in single-turn LLM fine-tuning. However, long-horizon agentic reasoning introduces increasingly branching interactions and sparse rewards, exposing several limitations of RL: its heavyweight backpropagation-based training stack makes it impractical to fine-tune larger LLMs, and longer-horizon trajectories make credit assignment in RL substantially harder. This paper argues that evolution strategies (ES) can be a better choice for fine-tuning long-horizon LLM agents. Compared with agentic RL, ES offers three key advantages: 1) Model Scalability: ES enables full-parameter optimization with only minimal, inference-level GPU memory, making it possible to fine-tune large LLMs. 2) Flexibility: its lightweight, black-box feedback interface makes ES fine-tuning easy to compose with prompt-space evolution (e.g., skill optimization & test-time compute); and 3) Long-Horizon Scalability: ES performs trajectory-level parameter attribution without decomposing rewards across horizons, yielding better scalability than Agentic RL as the horizon length grows. Based on this insight, we propose Agentic ESOpt, a full-parameter agentic fine-tuning framework tailored to flexible parameter--context co-evolution. At each step, Agentic ESOpt samples perturbations around the current LLM parameters, evaluates the resulting agents with rewards, and applies an online reward-weighted update. To improve the exploration--adaptation trade-off, Agentic ESOpt further introduces a cosine decay schedule of the perturbation scale $σ$. On WebArena-Lite, full-parameter optimization of Qwen-3.5-27B improves the No Skill baseline by 6.69%. In test-time automatic heuristic design, Agentic ESOpt performs online prompt--parameter co-evolution, improving its matched baseline in 28 of 36 settings.

While adversarial prompt tuning can enhance robustness of vision-language models efficiently, we find that existing methods aggravate robust generalization overfitting on seen classes, leading to a rapid degradation in performance against adversarial examples of unseen classes as training progresses. We empirically identify that this degradation stems from the tendency of the model to learn pseudo-robust features (i.e., non-generalizable shortcuts). To mitigate this, we propose ADAPT (Adversarial Disentangled Prompt Tuning), a robust prompt tuning framework following the philosophy of ``Learning What Not to Learn''. Specifically, ADAPT uses a dual-prompt mechanism with a target prompt and a pool of decoy prompts. During training, the decoy prompts are guided to entrap diverse pseudo-robust features, while the target prompt is constrained to be orthogonal to the decoys in the embedding space to learn robust features. By disentangling the robust features from the pseudo-robust features, ADAPT effectively prevents robust generalization overfitting. We further provide an analysis showing that the orthogonal loss bounds the effect of shifts in pseudo-robust features on unseen classes, yielding a testing error guarantee. Empirically, extensive experiments demonstrate that ADAPT substantially improves the robustness of the target prompt on unseen classes. The code is available at https://github.com/cheny02/ADAPT-ACMMM2026.

Formal verification is a crucial technique for ensuring the functional correctness of hardware designs. In the context of property checking, a key challenge is how to efficiently prove a user-specified property in the face of increasingly complex RTL designs. To address this challenge, abstraction techniques are often employed to reduce system complexity and accelerate the verification process. However, prior RTL abstraction methods either require significant manual effort or rely on rule-based techniques that lack flexibility. This paper introduces NeuroAbs, a neuro-symbolic framework for RTL abstraction. NeuroAbs first uses LLM-assisted RTL analysis to identify signals suitable for abstraction. It then combines LLM-based abstraction with an AST-based symbolic RTL representation to better align the generated abstraction with the intended transformation. The soundness of each abstraction is checked using satisfiability modulo theories (SMT) solving. If the abstraction is too coarse for a successful proof, NeuroAbs applies counterexample-guided abstraction refinement (CEGAR) to iteratively refine the model. Experimental results show that NeuroAbs significantly improves the efficiency of hardware property checking across a range of verification tasks.

Post-training with supervised chain-of-thought fine-tuning and reinforcement learning from verifiable rewards has substantially improved the mathematical reasoning capabilities of large language models (LLMs). However, their application to signal processing problems remains relatively under-explored. This report investigates reinforcement fine-tuning strategies for adapting Qwen2.5-3B-Base to graduate-level signal mathematical problems from WirelessMATHBench-XL, a comprehensive benchmark for mathematical reasoning in this domain. We examine two training paradigms: (i) direct reinforcement learning (RL) on WirelessMATHBench-XL with verifiable rewards; and (ii) supervised fine-tuning (SFT) on a distilled wireless-domain chain-of-thought corpus, followed by the same domain-specific RL stage. Across both paradigms, we benchmark Group Relative Policy Optimization (GRPO), Group Sequence Policy Optimization (GSPO), and Geometric-Mean Policy Optimization (GMPO). We aim to assess whether domain-aware CoT SFT serves as an effective initialization for subsequent RL, and whether GSPO or GMPO offer advantages in stability or accuracy over GRPO for signal reasoning tasks. Our best model achieves an overall accuracy of 39.12\%, representing a more than threefold improvement over the untrained Base model (12.37\%).

Time Series Foundation Models (TSFMs) have recently emerged as a highly promising paradigm for cross-domain zero-shot forecasting. However, existing evaluation protocols predominantly rely on static benchmarks with fixed historical test windows. While these benchmarks provide a valuable baseline snapshot, they evaluate an average performance on a fixed history, failing to capture how models behave in continuously evolving real-world environments characterized by seasonal variations, distribution shifts, and unexpected events. To bridge this gap, we introduce LiveHouse-TS, the first open-world living benchmark infrastructure for TSFMs. By evaluating models prequentially on real future data in open-world environments, LiveHouse-TS shifts time series benchmarking from snapshot accuracy to continuous temporal validity. Rather than acting as a one-off leaderboard, our infrastructure serves as a continuous time series infrastructure designed to explore vital, long-term scientific questions: Can model rankings be maintained over the long term? Which models remain genuinely robust under distribution shifts? Extensive streaming evaluations across 11 domains with 17 datasets demonstrate that static rankings undergo a dramatic reshuffling under a live protocol.

Existing research on irregular time-series forecasting has primarily focused on model design, while evaluation metrics remain insufficiently studied. Existing benchmarks typically use mean squared error (MSE) as the evaluation metric. We show that, in irregular forecasting, MSE is determined not only by the model prediction but also by the sample-specific timestamp sampling distributions, leading to a biased assessment of the models' continuous-time predictive performance. To address this issue, we propose the Continuous-time Squared Error (CSE), which employs importance weighting to eliminate the influence of the timestamp sampling distributions. We further theoretically prove that CSE's asymptotic estimation error with respect to continuous-time risk is no greater than that of MSE. Finally, we construct a systematic benchmark covering synthetic, semi-synthetic, and eight real-world datasets to validate the effectiveness of CSE and systematically evaluate models' continuous-time predictive performance. Experiments show that CSE can recover continuous-time risk more accurately than MSE, while relying solely on MSE may not fully reflect models' continuous-time predictive performance in real-world scenarios. Our code can be obtained at https://github.com/hnu-vis/ITS-Bench.

Group-relative policy optimization has emerged as a key paradigm for training agentic large language models (LLMs) on multi-turn interactive tasks. However, most existing variants fail to distinguish advantages among successful trajectories even when these trajectories differ substantially in their interaction efficiency. For instance, circuitous successes are often assigned the identical outcome reward, causing advantage collapse and severe performance bottlenecks. To this end, we propose Group Planning-aware Policy Optimization (PlanPO), a simple yet effective RL method for learning generalizable planning abilities beyond task-specific high-quality behavior patterns. Specifically, PlanPO introduces coarse-to-fine advantage signals, which capture the relative differences in trajectory-level lengths and turn-level response lengths conditioned on successful trajectories sampled for the same task. Within the group-relative optimization structure, this enables agents to actively learn generalizable and deliberate behaviors spanning interaction planning and textual generation from high-quality rollouts, without degenerating into vanilla length minimization. Experimentally, PlanPO improves over GRPO by 27.2\% on average across the challenging multi-turn benchmarks ALFWorld, WebShop, and SciWorld, outperforming recent powerful baselines while incurring negligible additional training cost.

GPT attention measures token compatibility through dot-product similarity. This mechanism is simple, effective, and memory-efficient. But it does not explicitly model whether strong token features should reinforce or suppress one another. We introduce Q-Interference, a fully classical quantum-inspired attention mechanism for autoregressive language modeling that augments each query and key feature with an amplitude and a learned phase. The resulting attention score is phase-aware which aligned phases contribute constructively while conflicting phases contribute destructively. Although Q-Interference yields a richer interaction rule than similarity alone, a naive implementation of Q-Interference requires a large token-pair-feature interaction tensor, making it memory-intensive and often impractical. To address this limitation, we propose an exact trigonometric factorization that computes the same score using two standard matrix multiplications avoiding materialization of the large intermediate tensor. Q-Interference fits directly into a Transformer block in GPT and leaves the remainder of the model architecture and next-token prediction objective unchanged. Experiments on public benchmark datasets and baseline models show that the proposed reformulation trains stably in a controlled GPT-style setting and provides a consistent memory advantage over naive phase-aware interference attention. These results support the specific contribution of this work: an exact memory-efficient reformulation that makes phase-aware interference attention practical within a standard GPT pipeline.

Compute-optimal scaling laws guide the training of frontier language models yet remain largely unexplored for visual generation. We present a systematic scaling law study for text-to-image diffusion models using Abra, a controlled family of flow-matching transformers trained across three orders of magnitude worth of compute ($10^{19}$ to $10^{22}$ FLOPs), reaching significantly larger compute budgets than previous works. We demonstrate that diffusion models scale just as predictably as language models but require far more data to train optimally: compute optimality occurs at approximately $200$ image tokens per parameter, ten times the Chinchilla compute-optimal prescription for LLMs. We show that unlike language models, diffusion models are robust to overtraining and that practitioners should err on the side of more data rather than a larger model. Finally, we show that this predictability extends beyond training loss to generative quality metrics, optimal CFG settings, representation quality, and even the shape of the training curves, which collapse onto a universal form.

Irregular time series forecasting is crucial in many domains, such as healthcare and meteorological observation. However, due to the inherent characteristics of irregular time series, including sparse observations and non-uniform sampling, accurately predicting future dynamics remains challenging. In light of these two characteristics, many existing methods aggregate irregular observations into fixed-dimensional estimated response coefficients through predefined basis functions and use these coefficients as sequence representations. Nevertheless, this modeling paradigm still suffers from two key limitations: (i) a potential non-vanishing asymptotic bias caused by ignoring the sampling density of timestamps; and (ii) the limited adaptability of predefined basis functions to diverse temporal patterns. In this study, we propose a Debiased Neural Basis-Function Network (DNBNet) to address these challenges. Its core is a debiased neural basis-function response mechanism, which corrects asymptotic bias through importance sampling while parameterizing basis functions with neural networks to adapt to diverse temporal patterns. In addition, considering the sparsity of irregular data, we design a novel multi-scale decomposition module based on average pooling, together with a mass-aware fusion mechanism, to obtain richer representations. Finally, a dual-branch decoder is employed for forecasting. Extensive experiments on multiple real-world datasets demonstrate the effectiveness of DNBNet and its strong generalizability across diverse irregular time series scenarios. Our code can be obtained at https://github.com/hnu-vis/DNBNet.

Existing agentic reasoning systems typically rely on centralized protocols. This design introduces routing bottlenecks and static role allocations that often fail when handling complex multimodal queries. We propose DeAR (Decentralized Agentic Reasoning), a framework that shifts from central control to autonomous peer-to-peer collaboration. DeAR is built on three mechanisms: (1) decentralized capability grounding for query-dependent agent specialization, (2) thought map navigation for targeted peer interactions, and (3) topology update for adaptive error correction. Evaluations across 9 diverse multimodal reasoning and text-based QA benchmarks indicate that DeAR consistently outperforms recent baseline methods, validating that decentralized and adaptive collaboration among agents enhances accuracy in knowledge-intensive reasoning tasks. The source code will be available at https://open_upon_acceptance.

AI agents increasingly act rather than merely read: across the Model Context Protocol (MCP) ecosystem, the share of deployed tools that modify external state has risen from 27% to 65% of tool use. When agents exercise this authority on public blockchains through MCP, skills, and tool calling, the consequences of an attack are governed by the blockchain execution layer rather than by conventional software assumptions. This survey argues that four properties of that layer (irreversibility, signing authority, continuous autonomy, and sequence-level composition) qualitatively change the threat model, turning the recoverable failures of generic agent security into a standing, irreversible loss. We organize the fragmented MCP-security literature into an attack-surface taxonomy, then contribute a Web3 risk-mapping matrix that ties each attack class to its amplified impact, the responsible amplifiers, a representative mitigation, and the residual gap. We synthesize defenses, including emerging blockchain-based mechanisms, and find them improving but insufficient: measured protections stop fewer than 30% of attacks, and model-level safety refuses fewer than 3%. We close by positioning the work against adjacent surveys and deriving a research agenda from the matrix's open cells.

Artificial superintelligence (ASI) requires AI to move beyond mastering existing knowledge toward exploring the unknown, creating new knowledge, and turning new ideas into verifiable results. However, the capabilities of today's AI systems are still largely built on learning, compressing, and applying existing human knowledge. Accordingly, existing benchmarks primarily test whether AI can produce correct answers based on learned knowledge, or whether it can complete tasks under extensive human guidance. We therefore introduce ASI-Bench, the first benchmark to jointly evaluate AI systems' capabilities of innovative exploration and autonomous scientific execution across general research domains, and the first to progressively withdraw human methodological guidance within the same research project to test how far AI can proceed on its own. Built by over 40 experts with the cost of 31,000+ human hours, ASI-Bench contains 60 project-level research tasks across 11 scientific domains and progressively reduces methodological guidance to test whether AI can independently select methods, conduct research, and produce verifiable results. All tasks undergo expert review, AI-assisted auditing, sandbox execution, and scorer validation. Across 18 state-of-the-art agent--model configurations, the average score drops from 50.91 with full methodological guidance to 29.10 with only the method specified and 26.62 when agents must determine the method themselves. This sharp decline shows that current systems remain heavily dependent on human guidance and are still far from autonomously conducting end-to-end, project-level scientific research. ASI-Bench is open to the world. We invite researchers and builders everywhere to contribute new tasks, challenge the limits of today's AI, and help accelerate humanity's collective path toward artificial superintelligence at https://asibench.apexin.ai/submit.

Large language models (LLMs) are increasingly used for scientific hypothesis generation. However, evaluating generated hypotheses remains a challenge for trustworthy AI-enabled scientific workflows. Existing approaches often use LLMs as judges or rely on semantic similarity, which can favor familiar ideas over novel ones. We propose a logit-based energy scoring method that evaluates hypotheses using a language model's intrinsic confidence rather than comparative judgment. We benchmarked seven language models on 1,323 papers across 12 disciplines. Each paper was paired with its hypothesis and fifteen incorrect alternatives. Intrinsic scoring reached 33.0% Hit@1 pooled across both scorers, compared with 16.6% for prompted listwise ranking. The strongest configuration, a 1-billion-parameter model using logit-based energy scoring, reached 53.1%, though this was the maximum across 14 model-by-scorer combinations selected post hoc. Overall, intrinsic model confidence shows potential for scientific hypothesis evaluation. This study also motivates future research on confidence-based methods for trustworthy AI-enabled scientific discovery.

Curriculum learning has been widely adopted in the post-training of large language models by organizing training data from easy to hard. However, its effectiveness varies substantially across reasoning tasks, suggesting that no single curriculum is universally optimal and raising a fundamental question: what determines when curriculum learning works? In this paper, we answer this question by analyzing the optimization dynamics induced by different curriculum schedules. We show that the transfer relationship between different difficulty levels characterizes the optimization dynamics induced by curriculum learning, which in turn explains the effectiveness of different curriculum schedules, and formalize this relationship as Relative Transfer, a principled measure of cross-difficulty knowledge transfer. Based on this measurement, we derive Transfer-aware Dynamic Curriculum Sampling (TDCS), which dynamically adjusts the sampling distribution according to the estimated transfer relationship throughout training. Extensive experiments on multiple reasoning benchmarks demonstrate that TDCS consistently outperforms representative scheduling strategies across different tasks, model scales, and training paradigms. More importantly, our work provides a unified optimization-based explanation of curriculum learning through cross-difficulty transfer.

This paper considers robust nonadaptive regulation for general nonlinear systems in an output-feedback setting with arbitrarily high relative degree. We develop a nonadaptive design that combines an input-driven filter and a generic internal model with a recursive backstepping law, thereby recasting the regulation problem as the robust input-to-state stabilization of an augmented error system. Unlike adaptive schemes, the proposed method does not rely on linearly parameterized regressors and does not require the construction of Lyapunov functions having merely nonpositive derivatives. Under standard assumptions on the exosystem, including purely imaginary and simple eigenvalues, together with a minimum-phase input-to-state stability condition on the internal dynamics, we establish global asymptotic regulation and derive explicit, verifiable inequalities for selecting the design gains. The resulting nonadaptive framework guarantees convergence of the estimation and tracking errors even when the controlled-system dynamics are complex or only partially known. The effectiveness of the theoretical results is demonstrated using a benchmark controlled Duffing system.

X-ray imaging can be approximately modeled as the projection of an underlying volumetric attenuation field, with each measurement recording the accumulated attenuation along a corresponding ray path. Reconstructing a CT volume from only a few X-ray views is therefore severely ill-posed, as the projections collapse depth information and leave 3D locations of anatomical regions and their corresponding intensity distributions highly entangled and ambiguous. We observe that once the spatial organization of anatomical regions is established, estimating their CT intensities becomes substantially more tractable. Motivated by this, we propose LiftXR, an interleaved, geometry-guided framework that explicitly incorporates spatial layout recovery into CT reconstruction. Specifically, a layout lifter first generates a 3D anatomical layout from bi-planar X-rays, providing spatial guidance for an intensity renderer to reconstruct a CT volume. An anatomical parser then performs volumetric perception on the reconstruction, exploiting its spatially resolved boundary and intensity cues to recover a refined anatomical layout. This transition from projection-conditioned layout generation to reconstruction-conditioned anatomical perception allows the parsed layout to provide feedback for region-specific intensity calibration. Extensive experiments on two public datasets demonstrate that LiftXR consistently outperforms recent X-ray-to-CT reconstruction methods, establishing a new state of the art. Moreover, the reconstructed CT achieves superior performance in external downstream segmentation, indicating improved anatomical fidelity. Code will be released.

Reinforcement learning (RL) has emerged as a powerful approach for improving reasoning in language and vision-language models, yet its strongest successes still depend heavily on ground-truth supervision (e.g., verifiable reward). Such annotations are costly to obtain and become increasingly scarce as reasoning capabilities advance beyond what humans can reliably evaluate. Self-rewarding RL reduces this dependence by enabling models to derive reward signals from their own completions. However, training solely on self-generated feedback can reinforce existing biases and suboptimal behaviors, reduce response diversity, and ultimately lead to homogenized responses and training collapse. In this work, we show that unsupervised reasoning can emerge through cooperative multi-agent training. We introduce Co-RL, a framework in which multiple decoupled models, sharing no parameters, are simultaneously optimized through RL using rewards derived from their peers. We further show that increasing cohort diversity, through heterogeneous model families, sizes, and rephrased training samples, reduces the correlated errors that drive self-reinforcing feedback loops. This diversity consistently improves reasoning performance, maintains behavioral diversity, and mitigates training collapse. Across text-only and multimodal domains, Co-RL consistently outperforms the base models and prior label-free approaches, while matching or surpassing supervised methods, without access to any ground-truth labels. Concretely, Co-RL yields average gains of 3.0-8.6% across seven text-only benchmarks for LLMs and 2.3-7.2% across four multimodal benchmarks for VLMs. Code is available at https://github.com/DrStranded/Co-RL.

This paper develops an adaptive surrogate modeling method for problems with very high-dimensional spatio-temporal outputs. The analysis of spatio-temporal multi-physics systems is computationally expensive and consists of a large number of inputs and outputs. Surrogate models are often constructed to replace the physics-based model to achieve computational efficiency in analyses such as uncertainty quantification and optimization that require many function calls. In order to address the challenge introduced by the high dimensionality of spatio-temporal output, a dimension reduction method is first employed to map the high-dimensional output to a low-dimensional latent space. This is followed by the construction of the surrogate model in the low-dimensional space. The prediction error in the original space, which includes both the reconstruction error and surrogate model error, is evaluated using different error metrics. Based on the prediction accuracy of the surrogate model, new training points are identified for adaptive improvement of the surrogate model. We present a novel adaptive sampling technique that combines exploration and exploitation to improve the surrogate model accuracy with the fewest possible runs of the expensive physics-based model. Thermo-mechanical analysis of a gas turbine engine blade is used to analyze the effectiveness of the proposed method.

When computational models (either physics-based or data-driven) are used for the sensitivity analysis of engineering systems, the sensitivity estimate is affected by the accuracy and uncertainty of the model. This paper considers global sensitivity analysis (GSA) for situations where both a physics-based model and experimental observations are available, and investigates physics-informed machine learning strategies to effectively combine the two sources of information in order to maximize the accuracy of the sensitivity estimate. Two representative machine learning (ML) techniques are considered, namely, deep neural networks (DNN) and Gaussian process (GP) modeling, and two strategies for incorporating physics knowledge within these techniques are investigated, namely: (i) incorporating loss functions in the ML models to enforce physics constraints, and (ii) pre-training and updating the ML model using simulation and experimental data respectively. Four different models are built for each type (DNN and GP), and the uncertainties in these models are included in the Sobol indices computation. The DNN-based models, with many degrees of freedom in terms of model parameters and training options, are found to result in smaller bounds on the sensitivity estimates when compared to the GP-based models. The proposed methods are illustrated for additive manufacturing and lake temperature modeling examples.

Personalized agents must decide whether retrieved user memory should be used, ignored, updated, or queried before it affects a current task. We use this setting to develop an empirical audit protocol for structured intermediate outputs: first audit dataset shortcuts, then isolate bundled prompt changes, check whether intermediate labels are answer-associated, test decomposed semantic evidence, and audit provider-level execution failures. A 480-example synthetic development set initially suggested large gains from a state-structured prompt bundle, but TF-IDF diagnostics showed lexical separability and no positive standalone Ignore cases. We therefore construct a frozen 160-example controlled counterfactual set with 40 matched four-way families and rule-derived reference policies. On this set, exposing the four state definitions improves accuracy, but an isolated explicit state-output field does not significantly improve policy accuracy for Llama-3.3-70B and gives only a marginal, non-significant gain for GPT-OSS-120B. Supplying benchmark-associated state labels shifts policy predictions, but because those labels deterministically map to policies, this is a label-conditioning diagnostic rather than evidence of a faithful internal mechanism. Family-level and seed-stability analyses further show that example-level accuracy overstates counterfactual consistency: complete four-way family success is rare. An exploratory follow-up that elicits decomposed semantic evidence also fails to improve routing for the cleanly evaluated endpoint; the corresponding GPT-OSS condition was unavailable because of provider-side request validation. We evaluate policy classification only, not downstream responses, tool actions, or memory-store mutation.

This article investigates several physics-informed and hybrid machine learning strategies that incorporate physics knowledge in experimental data-driven deep-learning models for predicting the bond quality and porosity of fused filament fabrication (FFF) parts. Three types of strategies are explored to incorporate physics constraints and multi-physics FFF simulation results into a deep neural network (DNN), thus ensuring consistency with physical laws: (1) incorporate physics constraints within the loss function of the DNN, (2) use physics model outputs as additional inputs to the DNN model, and (3) pre-train a DNN model with physics model input-output and then update it with experimental data. These strategies help to enforce a physically consistent relationship between bond quality and tensile strength, thus making porosity predictions physically meaningful. Eight different combinations of the above strategies are investigated. The results show how the combination of multiple strategies produces accurate machine learning models even with limited experimental data.

This paper addresses the limitations of Gaussian distribution assumptions in statistical sparse learning, particularly in modeling correlated and heterogeneous data. Conventional Gaussian models often lack robustness towards outliers and underlying distribution assumptions. To overcome these limitations, we propose the use of the $q$Gaussian distribution, derived from Tsallis entropy maximization, as a robust alternative. This is notably relevant in biostatistics, where the presence of correlated observations and heterogeneity, such as in genetic and longitudinal studies, is prevalent. Our contributions include modeling of correlated data through the re-derived multivariate probability density function from Tsallis entropy maximization, thereby addressing the limitations inherent in conventional Gaussian models. Furthermore, we introduce a novel framework that adapts numerical methods designed to find equilibria in flows to tackle composite optimization problems prevalent in statistical sparse learning. Applying this framework to the Hager-Zhang conjugate gradient algorithm \cite{Hager2005}, we develop a numerically stable and efficient algorithm for sparse statistical learning. The $q$Gaussian distribution, informed by the principle of maximizing Tsallis entropy, presents a viable and flexible alternative to Gaussian-based methods. This paper not only contributes to the theoretical understanding of statistical distributions and optimization techniques, but also paves the way for practical data analysis.

Converting structural framing plans into editable finite-element model drafts remains labor-intensive and prone to transcription error. Existing drawing-understanding systems for building components rely on task-specific trained neural detectors, and language-model agents in structural engineering operate on text or model data rather than the drawing itself. This paper presents, to the authors' knowledge, the first framework applying an agentic vision-language layer to structural component detection and model drafting from framing-plan PDFs, without task-specific detector training or fine-tuning. A deterministic stage extracts primitives, estimates scale by dimension-ratio consensus, recognizes five entity classes with a drafting grammar, and assembles an editable layout. The agentic stage proposes typed corrections constrained by deterministic candidates, operation-specific admission tests, change-level review, and fail-closed transactions. Evaluation used an author-generated benchmark of 100 plans: a development half that informed every rule revision, and a seed-disjoint held-out half generated after the rules froze, evaluated once. All reported scores are end-to-end results of the complete framework on the held-out half. Scale was estimated within 0.1% of the generator reference for every drawing. Recall and precision were 0.922/0.997 for columns, 0.886/0.990 for beams, 1.000/1.000 for walls, 1.000/1.000 for braces, and 1.000/0.964 for openings. A controlled study repeated two corruptions three times on three development drawings. Calibration passed all nine trials; member repair met every strict end-state predicate in five of nine. Guarded review corrected missed framing and false marks within explicit bounds. The held-out half shares the development generator, so the study excludes independently drafted plans, raster evaluation, analytical connectivity, and solver validation.

Multimodal large language models (MLLMs) are increasingly used to interact with screenshots, scanned documents, diagrams, and other visually grounded inputs. This shift introduces a new safety risk: in many multimodal jailbreaks, neither the prompt nor the image is harmful in isolation. Unsafe behavior emerges only when the model binds an apparently benign operation, such as summarizing, translating, or following, to a localized visual target. This reveals a structural weakness in current multimodal defenses, which largely moderate the prompt-image pair as a whole even though the true security-relevant unit is the grounded operation-target pair produced during dereference. In this work, we identify and analyze this reference-dependent failure mode and show that existing defenses degrade when harmful semantics are localized, activated only after grounding, and dependent on visual reference resolution. To address this problem, we propose COMIC (Context-Operation-Modality-Image-Classifier), a reference-aware pre-generation safety gate for MLLMs. COMIC first infers the requested operation and reference type, constructs candidate targets from OCR and open-vocabulary proposals, grounds plausible referents, and evaluates safety over explicit operation-target pairs. To handle ambiguity conservatively, COMIC combines max-risk aggregation with quality-aware routing before deciding whether to forward or block a request. We evaluate COMIC across multiple open-source MLLMs, localized and broader multimodal jailbreak benchmarks, and benign reference-sensitive settings. The results show that COMIC consistently improves robustness while preserving benign utility and practical efficiency. More broadly, our findings suggest that multimodal safety cannot be enforced reliably without modeling the requested operation, the visual target to which it applies, and the confidence of that grounding.

Low-cost, scalable screening for dementia remains an open problem. Imaging-based diagnosis is costly and hard to deploy widely. Electroencephalography (EEG) is portable and inexpensive, but its recordings are noisy, vary widely across subjects, and carry few clinical labels. We tackle this with Delta2Gamma, a self-supervised framework that learns EEG representations from unlabeled data by contrasting augmented views of each signal. Rather than treat EEG as a single stream, Delta2Gamma decomposes every recording into the five canonical neural rhythms (delta, theta, alpha, beta, gamma). Each band gets its own encoder and projection head. Each also gets a temperature that is predicted adaptively during contrastive training, so bands with different signal statistics are balanced automatically. On the ADFTD cohort under a strict leave-one-subject-out protocol, Delta2Gamma separates Alzheimer's disease from cognitively normal controls with 92.4\% accuracy. This exceeds both supervised backbones and recent dedicated EEG methods.

Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train-test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature-model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by $1.1\times$ to $6.5\times$, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio $1.75\times$). Conditioning on event type, mergers and acquisitions (M&A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC $= 0.138$ train-only, $0.068$ under train$\cup$val refit; 10,000-permutation $p < 10^{-3}$); the signal does not transfer to FNSPID's 2009-2020 U.S. corpus, localising the headline to our 2024-2025 European-tilted M&A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks.

Autonomous AI agents are emerging as interfaces for decentralized finance (DeFi) actions such as swaps, lending operations, and yield management. Because these agents rely on large language models (LLMs) to plan transactions, they inherit the LLM's susceptibility to prompt injection and lack of mechanisms to bind a verifier's approval to the exact transaction ultimately submitted on-chain. We present PACE (Policy-Attested Contract Execution), a transaction-level authorization framework that interposes between an LLM-based agent and on-chain execution. PACE introduces typed transaction intents, a deterministic policy verifier, and signed Policy Decision Records (PDRs) that cryptographically bind the approved intent, policy, and simulation report to the exact execution bytes, with replay and expiration protection. A Solidity smart account enforces PDR signatures on-chain with a measured overhead of 29,826-31,822 gas. We evaluate PACE against six baselines on 40 tasks spanning four attack categories plus benign utility (2,800 trials, 10 seeds). In our deterministic sandbox, PACE achieves a 0.00 unsafe execution rate and 0.00 false-positive rate on benign tasks, compared to 0.80 for the unguarded baseline. Ablation studies identify permissive policy settings (+57.5 pp) and the touched-contract allowlist (+12.5 pp) as the dominant safety components. To test whether the same deterministic floor holds for real model outputs, the artifact additionally provides a three-model live-LLM evaluation over the full task suite with repeated runs. A mainnet-fork harness is included for archive-RPC deployments, but fork results are reported only when the corresponding artifacts are generated. These auxiliary studies are separate from, and never substitute for, the deterministic benchmark. We frame our claims as logic-level safety within a reproducible benchmark rather than deployment-ready DeFi security.

Works of literature are complicated; they balance plot, suspense, surprise, and artistic expression. Summaries of literature prioritize plot, and therefore may deviate from their sources. Using a combination of manual and LLM-based annotation, we construct a dataset mapping sentences from 150 novel summaries to their respective source chapters. We find the task unexpectedly difficult for both human and model annotators. Using the sentence-to-chapter mappings, we then measure summary linearity, the degree to which it maintains the source's order of events, and uniformity, the degree to which a summary spreads attention equally across a source. By examining when and how summaries break linearity and uniformity, we identify differences in how literary works and summaries express plot, particularly with regard to the clarity and prominence with which narrative details are described.

A test oracle that obtains its expected value from the system it is judging cannot fail. If a fault moves measurement and expectation together the comparison cancels exactly, and no generated input will reveal it. The defect is in the oracle and not in the input space. We call this oracle anchoring. An expectation is specification-anchored when composed from values fixed outside the code under mutation, and state-anchored when it flows, directly or transitively, from that code. The expected-value form is named in the test-smell literature but not measured in any study we retrieved. We name three further channels by which such a value reaches a verdict, restrict the predicate to values flowing from the mutate target, and measure it. The subject is a deployed air traffic control simulator with 12 model-free property suites. Across 4 modules and 366 mutants these add 3 mutants of detection over the hand-written tests, while remaining 6 to 33 times as efficient per test. We then intervene three times, predicting each outcome first. Re-anchoring one holding oracle on published procedure, changing no production code, recovers 8 of 46; state-anchoring a healthy debounce oracle costs 4 of 19; and a reference model on that population kills exactly what specification anchoring kills, placing the risk in anchoring and not in model-freedom. Of 6 instances ablated, the two sizing their comparison carry 11 of the 12 recovered mutants. The published smell rule would revert our repair. Writing the oracle this analysis said was missing then exposed two defects deployment had not surfaced. All measurements come from one system by one author.

This paper challenges the pessimistic meta-inductive argument against scientific realism by undermining its inductive step rather than its historical premise. Although related challenges already exist, I develop a new one. Drawing on a general epistemology of scientific inference developed in frequentist statistics, machine learning, and formal epistemology, I evaluate induction in terms of convergence to the truth. I argue that ordinary enumerative induction can achieve everywhere convergence, whereas meta-induction fails even to achieve almost everywhere convergence. Indeed, in the problem context where meta-induction arises, the failure is deeper: no inference method whatsoever achieves almost everywhere convergence.

End-to-end vision-language-action (VLA) and world-action models offer an elegant route to general-purpose robotics, but their reliability is bounded by validated physical coverage. When an unfamiliar object, sensor, embodiment, or contact falls outside that coverage and no validated fallback exists, correcting the failure requires new robot data, a policy update, and regression testing. This recurring burden is the retraining tax. Unlike text, embodied data must often be created by operating machines. We present Teach-and-Grow Learning (TGL), an agent-centered architecture for general robot learning. In its general form, a multimodal agent turns a few successful demonstrations into reusable Skill Blocks: closed-loop behaviors for meaningful subgoals. In a new scene, the agent grounds and composes these blocks, selects learned or geometric tools, observes the physical outcome, and revises the route when execution departs from intent. A Skill Library stores executable behavior, while structured Experience Memory carries forward success, failure, and repair. New tasks are acquired without task-specific policy retraining. Our LIBERO evaluation attains state-of-the-art performance; controlled studies expose skill induction, persistent reuse, and agent-directed adaptation. Finally, we propose the Teach-and-Grow scaling-law hypothesis: if X denotes effective reusable experience, future-task error and teaching demand should approach irreducible floors as power laws in X. The architecture therefore treats deployment as a period of continued learning, in which one task can make the next easier.

Vision-language models (VLMs) combine images and text, but when the two conflict and one becomes harder to read, it is unclear how a model shifts its reliance between them. We study this modality reallocation with a controlled setup: we degrade either the image or the text across four levels of legibility while keeping the other clean, and track how the model's preference changes. We build conflicts from GSM8K and SVAMP by pairing the rendered image of one arithmetic problem with the text of another, so the two sources support different answers. We also introduce ChartQA-Conflict, a manually reviewed benchmark of 229 chart-report conflicts with matched chart and table-image representations. We evaluate six open-weight VLMs using both generated answers and a length-normalized conditional log-likelihood margin. On GSM8K and SVAMP, five of six models shift more strongly away from degraded text than from degraded images. On ChartQA-Conflict, all six likelihood-scored models exhibit the opposite pattern, shifting more strongly away from the degraded visual source. This reversal persists after calibrating for unimodal accuracy loss and after replacing charts with plain table images. Two frontier API models, GPT-5.6-Luna and Gemini-3.5-Flash, behaviorally replicate the ChartQA-Conflict reversal, with GPT-5.6-Luna also matching the arithmetic direction. These results show that modality reliance in VLMs is not fixed, but varies across tasks, evidence structures, models, and evaluation settings. The source code is available at https://github.com/Ro-netizen004/multimodal-arbitration-artifact.

Contrastive learning has become a cornerstone of modern representation learning, powering CLIP-style models that underpin text-to-image generation, vision-language models, and retrieval across a rapidly growing range of modalities. Despite this empirical success, the expressive power of these architectures remains poorly understood. To gain insight, we study expressivity by adopting a population-level, density-estimation viewpoint: each architecture comprises a parameterized set of densities whose parameters may be chosen to approximate the joint distribution of the modalities. This isolates a question of pure representational capacity: which joint distributions can a given contrastive family of parameterizations approximate to arbitrary accuracy? We show that expressivity is sharply architecture-dependent. For two modalities, the simple two-tower CLIP architecture is a universal approximator. A natural generalization of CLIP, widely used in practice when three or more modalities are present, is based on a loss found by summing over all pairwise similarities. This provably cannot represent arbitrary joint distributions, although we prove that it remains expressive enough to match all pairwise conditionals. Motivated by this gap, we propose Hadamard-CLIP, which adds a single learned weight vector on top of the existing encoders and restores universal approximation of the joint for any number of modalities while preserving CLIP's fast, precomputable-embedding retrieval.

Safety alignment in open-weight language models is trivially removable: abliteration projects a refusal-mediating direction out of the weights in minutes, and no release-time defense we are aware of prevents it durably. What cannot be prevented can be deceived. Our defense, decoy hardening ("Fool's Gold"), concedes the refusal strip and poisons its payoff: once refusal is stripped, most answers to hazardous operational requests are confident, fluent decoys whose critical elements are falsified. Decoys are trained inside a differentiable simulation of the attack, expressing only in the attacked state; a refusal pin and benign leash hold clean-state behavior to the original. We instantiate it on seven models from five families (9B-122B, dense and mixture-of-experts). On the six models passing our pre-registered efficacy gate, 0.51-0.90 of attacked-state responses to held-out prompts are decoys, +0.27-0.84 attributable to the defense; all six stay within registered benign-behavior and capability budgets; the seventh (smaller) fails the gate (boundary case). Rates replicate on a frozen test split or untouched strata. The claim is epistemic: without independent ground truth, no observation surface we tested separates falsified answers from correct ones - on external red-team benchmarks' CBRNE-adjacent slice, the defended 122B is fatally wrong on 0.82-0.86 of matched-quality answers vs at most 0.10 undefended. Repeated sampling does not restore trust: element-wise consensus at K=64 reconstructs a fully usable procedure on 0.083-0.625 of prompts where the instrument validates, vs 0.58-0.96 undefended, with no label-free way to tell the regimes apart; on the weakest such model the claim is per-draw only. We evaluate chemical and biological hazards; the defense does not address in-context jailbreaks and protects only the initially released defended weights.

We present Graphectory Viewer, a web-based tool for interactive, process-centric analysis of software-agent trajectories. Building on the Graphectory representation introduced in our previous work, Graphectory Viewer transforms heterogeneous raw trajectories into phase-aware graphs that connect low-level execution details with higher-level behavioral structures. The tool supports trajectories from multiple agent frameworks and provides interactive graph construction; node-level inspection of thoughts, actions, and observations; search and filtering over large trajectory collections; and Sankey-style summaries of problem-solving phase transitions. These capabilities enable researchers and practitioners to inspect individual executions, identify recurring behavioral patterns, compare successful and failed runs, and analyze large trajectory corpora beyond final task outcomes. To support reproducibility and further research, we release Graphectory Viewer as an open-source artifact together with documentation, precomputed graphs, and the large-scale trajectory corpus.

Dimensionality reduction methods are instrumental to visualize high-dimensional data, and t-SNE stands as one of the most widely used methods due to its emphasis on local neighborhood preservation. A central component of t-SNE is the affinity matrix, which expresses pairwise similarities in the form of symmetrized probabilities, over which the optimization problem of t-SNE is defined. We study how the sharpness of this probability distribution affects neighborhood preservation at different scales. We introduce a row-wise power transform controlled by a parameter gamma that can smooth or sharpen each row of the affinity matrix while preserving sparsity and rank order. We show that this transform is equivalent to rescaling the Gaussian bandwidth and thus to changing the perplexity. However, as the sharpness of the probability distribution varies per point, a fixed gamma leads to point-dependent effective perplexities, making it distinct from changing the global perplexity. Empirically, we find that sharpening improves preservation of the very nearest neighbors, while smoothing improves preservation of broader local neighborhoods, outperforming alternative affinity constructions including multiscale methods in the mid-local range.

Multi-agent AI workflows are limited not only by model quality but by token cost, latency, and context-window quality. This paper presents a practitioner framework for token optimization and context-window management, grounded in an internal production dashboard that extracts structured work items from meetings, email, and chat with LLMs and routes summaries across workstreams. Six patterns are described: context stratification, fetch-once/process-locally architecture, schema-contracted prompts, token-aware fallback chains, semantic caching, and inter-agent communication compression. In production they cut measured cold-load latency to 61-116 seconds (six timed runs) from an operational baseline of roughly 3.5-10.5 minutes, with an estimated 60-70% token reduction. It also reports a controlled context-composition study: 2,420 confirmatory trials across 11 model configurations, using 661 anonymized workplace items scored for relevance. Holding the prompt at a fixed ten items, replacing some high-relevance items with same-domain low-relevance items improves the model's relevance-score concordance on the target items, versus high-relevance items only; we call this relevance-contrast context. In the all-11 paired analysis, the 50:50 signal/noise condition improved relevance accuracy by +0.077 over the 100% condition (naive 95% CI [+0.056, +0.098], Cohen's d = 0.49, Holm-adjusted p < .001, n = 220). These cells are not independent; by the nine model families the effect is +0.084 (95% interval [+0.064, +0.103]), reported as a within-corpus descriptive comparison, not a population inference. A Fusion-of-N follow-up found that learned synthesis did not beat the mechanical set union of item IDs. The contribution is a measured engineering layer between model research and production agent practice: repeatable patterns and evaluation methods for faster, cheaper, more reliable workflows.

Job Safety Analysis (JSA) and pre-task planning can benefit from prior incident records, yet historical accident data is often stored as unstructured narratives that are difficult to consult at the point of planning. A novel framework centered on large language models (LLMs) for highway construction safety reporting and planning is proposed as a foundation for future agentic applications, prioritizing deterministic, local inferencing. The first aim is to enable classification and quality scoring of incident narratives for existing and future reporting purposes. The second is to evaluate retrieval of relevant historical accidents, related imagery, and trusted industry documents for incorporation into daily safety plans. Neural probes were trained to classify incidents along four multiclass and two binary Occupational Injury and Illness Classification System (OIICS) fields and to derive an overall quality score, evaluated on a test set of over 15,000 narratives and a held-out set of 100 author-labeled records, benchmarked against a majority-vote LLM ensemble. The retrieval of historical accidents, reference imagery, and industry documents was benchmarked across embedding models using standard information retrieval metrics. OIICS classification reached 75% held-out accuracy, though the two binary flags were degenerate. The quality score, while meaningful on one database, was distorted on out-of-distribution fatalities in the held-out dataset. Accident retrieval recovered relevant incidents far above chance, performing best on lexically distinct construction activities. On document question answering, an open-weight decoder embedding model surpassed proprietary models. Overall, this work provides a new framework rooted in local inferencing and text embedding models for future agentic applications, with emphasis on bridging external data to JSA reports.

Small Language Models (SLMs) are increasingly deployed in resource-constrained, privacy-sensitive settings, where safety and bias failures can cause security and societal risks. However, existing AI safety\slash security\slash compliance benchmarks are designed for large language models that may not transfer reliably to SLMs. We therefore ask: Can these benchmarks effectively and reliably evaluate SLMs? To answer this question, we conduct a large-scale assessment of the effectiveness and robustness of these automated pipelines by evaluating five widely used benchmark suites across 26 open-source SLMs under a unified judging rubric, which assigns a score of 0, 1, or 0.5 to harmful, safe, or ambiguous/irrelevant responses, respectively. Across the benchmarks, ambiguous judgments dominate and correlate with prompt complexity and model architecture, indicating that {\em LLM-centric safety benchmarks are insufficient as standalone evidence for SLM safety assessment}. In general, the ambiguity rate increases with lexical density, output perplexity, and output length and decreases with lexical sophistication, self-coherence, and reply-prompt similarity. This reveals a capability-safety confound that mixes model capability with apparent safety. Since ambiguity is prevalent, aggregate mean-score leaderboards are mathematically brittle: model rankings change significantly under reasonable ambiguity treatments, even when the underlying outputs remain unchanged.

Reinforcement learning (RL) theory fundamentally depends on probability theory through the Markov chain. There is a deep connection between probability theory and potential theory. This paper reviews that connection and explores the potential-theoretic viewpoint for core reinforcement learning representations and algorithms under a fixed-policy assumption. This viewpoint may offer a path for improved sample efficiency and formal constraints that can be applied to RL. When the fixed-policy assumption is relaxed, the linear potential theory framework can be naturally extended to the nonlinear case.

Contextual Reinforcement Learning (CRL) seeks to generalize classical RL by maximizing task coverage across a context space of related tasks. While prior works often train from scratch and rely on either multi-task learning for a single policy or strategically training multiple policies, we advocate for a unified alternative: pretraining a single policy with good initial performance, followed by fine-tuning multiple policies for task specialization. This new paradigm, however, introduces unique challenges, such as heterogeneous marginal returns and sample inefficiency. This raises a critical research question: given a pretrained policy and a constrained budget, how much fine-tuning should each task region receive to enable sample-efficient CRL? To this end, we propose Task Specialization Fine-Tuning (TSFT), an online framework that predicts fine-tuning performance with a simple parametric model and exactly solves the resulting discrete budget allocation problem via integer linear programming. Extensive experiments across diverse decision domains, including combinatorial optimization, continuous control, and LLM fine-tuning, demonstrate that TSFT significantly outperforms baselines in task coverage and approaches oracle performance. Our work charts a new direction for model-based CRL, aligning with the modern pretrain-finetune era.

LLM-based agents are increasingly used for coding tasks, where they have outperformed many classical approaches and scaled to repository-level tasks, such as test generation. However, when directly prompted to generate tests, these agents can fail to reason about the code and its underlying contracts, thereby missing edge cases and behavioral boundaries that affect test quality. To address this limitation, we propose Spec-Driven Test Generation, where we instruct an agent to first reason about -- and explicitly document -- code pre-conditions, post-conditions, and undefined behaviors. This intermediate semi-formal specification acts as a cognitive scaffold to guide subsequent test generation. Our evaluation on production bugs from Google shows that the spec-driven agent can deliver a 9.8 percentage points ($p = 0.0352$) improvement in bug detection rate and a 2.5 percentage point ($p = 0.0034$) improvement in branch coverage, compared to a traditional test generation agent baseline. Using LLM-as-a-Judge, we further show that test suites generated by the spec-driven agent are superior to the baseline and human-authored tests in 77.8% and 56.7% of the cases, respectively, and demonstrated improvements on following best practices, readability, and edge-case coverage.

An AI audit record is useful only if its durability and trust boundary are explicit. Returning a guarded decision before any durable write minimizes latency, but it cannot guarantee that evidence survives an immediate crash. We rebuild RuntimeGuard-AI around this constraint. The resulting research prototype binds each deterministic policy decision to the exact policy source, commits a privacy-minimizing record at a caller-selected synchronization boundary, and returns an Ed25519-signed receipt that states whether that boundary completed. After restart, the engine validates framed records, manifests, shard placement, sequence continuity, and replay identity. A separate attestation path groups committed records into chained, signed Merkle epochs that an auditor verifies with an externally obtained key. On an Apple M4 Pro at four worker threads and 2,048-byte prompts, buffered signed evidence reaches 27,193 requests/s with 141.9 microseconds median latency. Per-record data and full synchronization reduce throughput to approximately 242 requests/s and raise median latency to 16.0 ms. Sealing a 100,000-record signed epoch takes 97.0 ms. The result is a measured durability-latency trade-off, not a "free" asynchronous audit path. The prototype does not prove model execution, prevent a compromised signer from forking history, or establish legal conformity.

Chronic kidney disease (CKD) progresses silently and severely undermines quality of life, making early detection critical for improving patient outcomes. We present a two-part study that combines large-scale telehealth data with advanced machine learning to both classify self-reported CKD status and identify key drivers of disease. Using selected features from the Behavioral Risk Factor Surveillance System (BRFSS 2021: 438,693 samples; BRFSS 2019: 418,268 samples) and the National Health Interview Survey (NHIS 2021: 29,482 samples; NHIS 2020: 31,568 samples), we addressed missing data with nine state-of-the-art imputation methods and mitigated class imbalance via sampling strategies. Our customized stacked ensemble model achieved balanced accuracy of 72.56-76.12%, with corresponding AUROC scores of 79.59-82.29%. SHapley Additive exPlanations (SHAP) analysis, followed by clinical review, highlighted critical predictors, including regular medical check-ups, age, blood pressure, and indicators of mental health stress. These findings deliver a robust and interpretable framework for CKD risk stratification and provide actionable insights into its associated factors.

Online decision making often requires navigating a landscape shaped by both dynamic contexts and strategic interactions. In competitive pricing, for example, hotels must account for both dynamic contextual factors and rivals' strategic responses. Existing approaches address only part of this challenge: contextual bandits optimize single-agent decisions using observable features but ignore multi-player interactions, while online matrix games capture strategic behavior through Nash equilibrium but assume fixed payoffs, ignoring contextual information. How should agents act then when strategic payoffs evolve with contextual signals? We introduce \emph{online contextual matrix games} to integrate contextual information into multi-player online games. We further propose \emph{OnGameLearn}, an online learning algorithm that efficiently balances exploration and exploitation across both player actions and contexts. This approach comes with statistical guarantees: tail bounds for the estimated payoff matrix, the convergence of the estimated Nash equilibrium, the asymptotic normality of the parameter estimators, and the sublinear regret bound. We also develop the notion of \emph{policy value} in matrix games and develop a doubly robust, $\sqrt{T}$-consistent estimator for it. Across simulated studies and a real-world hotel pricing application, we find that OnGameLearn effectively navigates the intertwined challenges of strategic and contextual decision-making.

Many table-centric NLP tasks such as NL2SQL first retrieve relevant tables from large collections using keyword search. Recent work uses LLMs to generate natural-language table descriptions to improve retrieval, but they are typically optimized for fluency rather than retrieval effectiveness. We present Polaris, a system that trains an LLM to generate table descriptions directly from retrieval feedback. Our key insight is that existing table retrieval benchmarks already contain the supervision needed for this task: given query-table relevance judgments, we generate multiple candidate descriptions for each table, rank them by their BM25 retrieval effectiveness, and use the resulting preference pairs to fine-tune the LLM with Direct Preference Optimization (DPO). Polaris further expands abbreviated table and column names before generation to reduce vocabulary mismatch. Extensive experiments show that Polaris outperforms the state-of-the-art AutoDDG solution, often by a significant margin. More broadly, our results demonstrate that retrieval benchmarks can be repurposed as supervision for training LLMs to generate retrieval-oriented metadata.

Algorithm selection for constraint satisfaction problems requires extracting features that capture problem structure. Manually designing feature extractors demands deep domain expertise and quickly becomes a bottleneck when new problem classes appear. We present an automated approach that uses Large Language Models (LLMs) in an agentic check--fix--verify loop to synthesize executable Python scripts that act as interpretable, problem-specific feature extractors. Given a high-level MiniZinc model and an instance, the LLM agent generates code that constructs a typed graph representation and computes structural properties such as graph density, variable clustering, and constraint tightness. We evaluate our approach on three combinatorial problems (vehicle routing, car sequencing, fixed-length error-correcting codes) with a portfolio of five state-of-the-art solvers. The synthesized extractors yield algorithm selectors that consistently outperform both expert-curated mzn2feat features (up to $8.3$ percentage points (pp) test-set accuracy on FLECC) and the best transformer-based trans2feat variants. In the meanwhile, the synthesized feature extractors remain inspectable.

Reasoning has become a standard technique and feature for contemporary LLMs; however, its application and quality in the context of demanding legal-oriented tasks, such as legal case forecasting, remain under explored. We investigate how LLMs reason in the context of legal case forecasting, using legal cases from the European Court of Human Rights (ECtHR) as a testbed. We evaluate OpenAI GPT 5.4, a recent top-tier LLM, by exploring alternative prompting strategies that are more or less suggestive of what counts as legally meaningful reasoning in the context of ECtHR jurisprudence. We present our findings derived from assessing the model's responses with both human and LLM evaluation. We find that the examined model scores far from ideal in legal reasoning, the model produces structurally complete but substantively shallow analyses, and that LLM-as-a-Judge evaluators are internally consistent yet align only weakly with our trained annotators, i.e., reliable but not a valid substitute for human evaluation. Overall, the expert-curated prompt leads to more comprehensive reasoning, which does not result in more accurate predictions compared to the other examined settings. Based on our findings, we urge the community not to rely solely on automated LLM-based evaluation and to avoid using task accuracy as an appropriate proxy for reasoning quality.

Active inference selects actions by minimising an expected free energy functional over predicted futures. However, adding an expectation over yet-unobserved outcomes means the free energy functional no longer has a Kullback-Leibler structure, which hinders message passing treatments of inference procedures. We propose an alternative formulation based on a Bethe free energy functional, fully supporting inference by message passing. The epistemic drive is maintained by imposing an information constraint, next to normalisation, marginalisation and form constraints, insisting that the mutual information between future observations, states and parameters given actions must be at least as large as the entropy of the goal prior. For a specific value of the corresponding Karush-Kuhn-Tucker multiplier, the stationary point of this constrained Bethe Lagrangian recovers the expected free energy solution. We show that, as the information demand is varied, the solved multiplier moves through its inactive, interior, and saturated regimes. In the inactive regime the agent's epistemic drive switches off entirely, while in the saturated regime it is maximal. We compare the performance of the constrained Bethe agent on three tasks against EFE and Q-MDP.

Textual context such as news, reports, and logs can provide valuable signals for time series forecasting, especially when future dynamics are driven by external events that are not yet visible in historical values. Existing multimodal forecasting methods often either ask large language models (LLMs) to predict numerical values directly or fuse text and time series implicitly, making contextual influence difficult to interpret and control. We propose SCENARIODIFF, a hierarchical contextual reasoning framework for multimodal time series forecasting under noisy and weakly aligned documents. SCENARIODIFF organizes contextual information into three levels: a Historical Context Agent extracts stepwise evidence from raw documents, a Scenario Agent produces a qualitative scenario description for the forecast horizon, and an Anchor Guidance Agent generates sparse anchor points for event-relevant future regions. These structured signals condition a Multimodal Diffusion Transformer, while Anchor Blended Sampling locally refines generated trajectories without retraining. Experiments on the Time-MMD benchmark show that SCENARIODIFF is especially effective in event-driven domains, demonstrating the value of explicit hierarchical scenario guidance for multimodal time series forecasting. Our full implementation is available at https://anonymous.4open.science/r/ScenarioDiff_ICDM-2C4C

Off-policy reinforcement learning (RL) has become increasingly sample-efficient, enabling applications such as RL fine-tuning of Vision-Language-Action models into reliable, high-performing policies. World models offer a further lever for sample efficiency, as they predict state changes rather than actions alone, but their success has largely been confined to supervised policy learning. Prior model-based RL methods often optimize the policy or value function directly on imagined rollouts, which is prone to compounding bias and struggles to scale to large, high-dimensional problems such as real-world robotics, a problem that worsens with task horizon and visual complexity. In this work, we instead ask whether we can leverage world models directly on top of standard Q-learning to improve performance, while remaining trained and grounded in the real, online setting. We propose QWM, a framework that leverages world models to perform test-time search over imagined trajectories on top of Q-learning to select high-value actions during both online rollouts and evaluation. Since the policy and value function are trained only on real transitions, QWM avoids compounding model bias while still gaining the sample-efficiency benefits of predictive search. On challenging manipulation benchmarks Robomimic and LIBERO, QWM significantly outperforms strong prior state-of-the-art methods on both sample efficiency and performance.

What a language model internalizes from fine-tuning is usually diagnosed after the fact. We make it an experimental variable. OraclePhys is a systematic fine-tuning framework with three components: OraclePhys-Bench, an exactly-graded structural-mechanics benchmark whose finite-element oracle scores every answer and counterfactual edit -- no human labels, no LLM judging; OraclePhys-30K, a supervision dataset of seven answer forms over byte-identical structure descriptions; and a controlled training study across the seven forms and three verifier roles. The study yields two findings. First, the label's answer form -- not its bit count -- causally determines what fine-tuning teaches: a ranking objective installs an out-of-distribution forward model where the untrained base sits at the guessing prior, a scalar objective at best a partial one, a boolean nothing detectable; the vector-scalar gulf survives a second physics domain, a second model family, and a paraphrased evaluation surface. Second, written or score-filtered answers install this capability, while advantage-weighted scores (GRPO) raise reward yet leave the model statistically equivalent to its start on held-out physics -- within the recipes and budgets tested -- sufficing only for routing. The trained 8B -- the first LLM on spatial structural response -- reaches the task's data-precision frontier: above a frontier LLM at zero- and 32-shot, at a specialist's level. What the label spells out about the target computation is what fine-tuning teaches; what you train on is what you route.

Research software projects describe themselves in many places at once: citation files in the repository, archive deposits, DOI registry records, package registries, and README text. We treat the software as the underlying object and these machine-readable self-descriptions as its surfaces: the points where people and automated systems read what the project declares about the software. Citation guidance, indexing services, and automated agents may read a different subset of these surfaces, so disagreement between them can silently fragment credit and provenance. This paper asks a simple question that has not been measured directly: when a project's own metadata surfaces are compared with each other, how often do they agree? We audited 117 open-source research software projects, comprising an 87-project high-performance computing and quantum computing corpus and a 30-project registered baseline drawn from the JOSS and pyOpenSci accepted-package lists, across up to seven machine-readable surfaces per project. Using a four-level verdict rubric across six metadata fields, with 98.5\% hand-verified verdict precision on a 338-row stratified sample, we found that 52 of the 62 projects exposing at least two comparable surfaces (83.9\%) contain at least one core-field conflict, a result that is insensitive to the fuzzy-matching threshold. Half of hand-adjudicated cross-surface conflicts trace to a single mechanism: surfaces describing the software's paper rather than the software itself. Among projects whose CITATION.cff includes a preferred citation, 28 of 32 route citations to a record that disagrees with the software's own metadata. The author lists and titles disagree the most, and the registry surfaces are the least aligned. We release the audit pipeline as an importable library, the corpus, the registered sampling protocol, all raw snapshots, and the complete verification log.

Retrieval-Augmented Generation (RAG) has significantly enhanced the performance of large language models (LLMs), yet these systems remain vulnerable to knowledge-poisoning attacks, in which misinformation in retrieved documents can influence the model's final outputs. Notably, an LLM may correctly detect that a document contains incorrect information while nevertheless being influenced by it. Prior work has addressed this vulnerability through the Cordon Principle, which prevents models responsible for final answer synthesis from directly accessing raw evidence. Although effective, this strict isolation can introduce substantial computational overhead. In this work, we propose a refined security principle: only agents capable of deliberative System 2 reasoning may access untrusted documents. To evaluate this principle, we introduce novel metrics that quantify the discrepancy between misinformation detection and downstream influence. We then empirically compare state-of-the-art reasoning language models with standard language models across these metrics. Our results show that reasoning-capable models are substantially more robust to corrupted evidence, without requiring the strict isolation imposed by the Cordon Principle. These findings provide empirical support for our refined principle and suggest a more practical foundation for secure RAG system design.

Cell mimicry arises when different cell types appear morphologically similar. Human pathologists resolve this ambiguity using surrounding tissue context, whereas current vision models either lack contextual reasoning (cell foundation models) or cannot operate at the cell level (pathology MLLMs). We present Loki-OT, which propagates region-level tissue reasoning to individual cell predictions via Unbalanced Optimal Transport, using MLLM-derived density priors as soft guidance for ambiguous cell reassignment. Loki-OT is motivated by the observation that pretrained cell foundation model features already encode discriminative information, including tissue context, but standard cell-level supervision fails to use tissue context effectively. The resulting transport plan is distilled into a lightweight student MLP classifier that learns context-aware decision boundaries within the pretrained feature space. On the independent TCGA-BRCA cohort, Loki-OT achieved lower patient-level MAE than the fully supervised in-domain PanopTILs classifier and improved F1 in epithelium-rich mimicry tissues, using 278 weak region-level MLLM estimates built on a general-domain cell foundation model. Code: https://github.com/xiangli980/Lymphocyte_Mimicry_Correction_via_Loki_OT

To effectively collaborate with users on knowledge-intensive tasks, Large Language Models (LLMs) must perform information calibration: matching content to a user's evolving understanding and cognitive capacity. Yet user simulators used to evaluate and train LLMs do not explicitly model user knowledge so they neither produce realistic interactions across knowledge levels nor reflect how interactions unfold as that knowledge evolves. To close this gap, we introduce KNOWSIM, an evaluation framework built around a user simulator that maintains explicit knowledge states, represented as a graph of Information Units with prerequisite relationships, that evolve under update rules grounded in learning theory. KNOWSIM computes three metrics (Knowledge Gain, Delivery Calibration, Cognitive Overload) directly from the knowledge state trajectory, reflecting key mechanistic aspects of information calibration. We validate KNOWSIM against 705 human-AI sessions across two domains, stratified by knowledge level: its rankings align significantly with human judgments (73-74% sign agreement), outperforming three baseline simulators. Applied to 9 LLMs, KNOWSIM reveals that the best model shifts by user knowledge level, revealing aptitude-treatment interactions invisible to standard evaluation.

A personal language agent learns a fact from one audience and may later place it in the prompt it assembles for another. This memory-to-context step is an attack surface: ambiguous or inconsistent channels, cross-audience prying, and poisoned memory can each cause the system to assemble context containing a fact relevant to the query yet unauthorized for the current viewers. We introduce authorization before context: a single, anti-monotone audience-membership rule applied at the memory-to-context transition. Each item carries the audience present when it was recorded; the current viewer set is read from channel metadata and falls back to public when ambiguous; and the item is admitted only when every current viewer already belonged to its audience. We prove that this rule gives every participant cross-channel recall while ensuring, by exclusion rather than by model behavior, that nothing recorded for a narrower audience reaches a broader one and that poisoned memory cannot widen its own audience. The boundary is a model-neutral invariant on the exact assembled context: a forbidden fact must be absent before the model is called. On a synthetic Contextual-Integrity suite, no forbidden fact entered the context our boundary assembled, whereas unscoped baselines included such facts by construction; we further audit that every read path fails closed. The evidence is preliminary and synthetic.

Image-to-image face generators are widely used, and visual dissimilarity between their outputs and source images is sometimes treated as evidence of privacy. Auditing whether these systems satisfy formal identity-level (epsilon, delta)-differential privacy requires choosing among several distinct routes for converting embedding-space observations into estimates or bounds on the differential privacy parameter epsilon. We present a comparative study of four such audits applicable to pre-trained, black-box face generators: a Gaussian-mechanism reading of per-identity sensitivity (GaussMech); a per-dimension kernel-density log-ratio aggregated by basic composition (KDE-LR); an analytical population-level lower bound on pure-DP epsilon derived from the maximum mean discrepancy via the total variation distance (MMD-TV); and a hypothesis-testing evaluation of a cross-validated classifier's out-of-fold ROC (ROC-HT). For each method we make explicit its assumptions, hyperparameter dependence, finite-sample limitations, and the regime in which its epsilon estimate is informative. Applied to FaceFusion and InstantID across multiple identity encoders and reference datasets, the audits consistently reveal substantial identity distinguishability while reporting markedly different epsilon estimates that reflect each method's distinct assumptions and finite-sample treatment. In this high-distinguishability regime, the experiments do not support a reliable ranking of the four methods. Their relative trade-offs should be evaluated on partially private mechanisms, which we identify as the natural next study. The resulting framework places these audits in a shared identity-level audit setting and clarifies how their assumptions and finite-sample treatments shape the resulting differential privacy estimates.

Tensor networks are powerful formats for compressing large-scale data. However, their application to general data processing has been limited by the difficulty of performing nonlinear operations. Here, we introduce iterative tensor network transformations (ITNTs), a general algorithmic framework for the element-wise evaluation of elementary and nonlinear filtering functions on data encoded as tensor trains (TTs), a class of tensor networks. Our approach operates entirely in the compressed domain, enabling efficient computation on exponentially large datasets while maintaining a controlled computational cost. We demonstrate its power in two key areas: (I) evaluating highly nonlinear elementary and filtering functions on a 3D reactive flow field, enabling high-fidelity reaction rate computation and region filtering, and (II) finding extrema in complex optimization problems, such as solving Max-SAT instances on spaces up to $2^{70}$ configurations. These results establish ITNT as a foundational tool that provides tensor network methods with the capability for general-purpose data science and large-scale optimization.

Recovering the directed acyclic graph (DAG) of a structural equation model (SEM) from observational data is a central problem in causal discovery. The iterative gradient descent and per-problem hyperparameter tuning of continuous-optimization methods are poorly suited to two practically important regimes: the sample-limited regime, where the number of samples is comparable to or smaller than the number of nodes in the DAG, and the compute-limited regime. This work proposes SURE-Ridge, a non-iterative, closed-form estimator for equal variance linear Gaussian SEM. The method performs parallel node-wise regressions with regularization parameters chosen adaptively by Stein's unbiased risk estimate (SURE), and applies an adaptive thresholding procedure to extract a DAG from the resulting soft adjacency matrix. Numerical results show that SURE-Ridge achieves the lowest structural Hamming distance in the small-sample regime and the lowest run time across all sample sizes tested, compared with NOTEARS, DAGMA, and GBNSL baselines.

A personal AI system needs a model of the user's goals, constraints, and ongoing commitments to plan and act on their behalf, and the quality of that model is bounded by what the system can observe. Broader observation does not by itself improve assistance because a bounded system must select and compress information for the task at hand. We argue that this observation bottleneck has a cooperative structure: the system builds a partial model of the user's changing life, the user evaluates its actions, and the user's consent and control shape what it can observe next. Useful and inspectable behavior can give users a reason to maintain or expand the observation channel, while failures can lead them to correct, narrow, revoke, or abandon it. We use the term cooperative observation for this feedback loop among usefulness, trust, and future access, and propose it as a framework for personal intelligence. We report a preliminary single-subject account from Organizm, a prototype used over six months, and outline evaluation directions for measuring how observation quality shapes personal AI.

Combining the answers a large language model (LLM) samples for a question into one decision is a test-time information fusion problem, usually solved by majority voting. Voting is unreliable on difficult questions, where the sampled answers share correlated errors, so the wrong answer can win and drawing more samples makes the decision worse. Selecting a candidate by reading a correctness signal from the model's hidden states is a promising alternative, but its accuracy varies across models and tasks, and no measure indicates when it can be trusted. In this paper, we propose CASE (Correctness-Axis SElection), a dynamic selection combiner that trains a linear gate on the answer-token hidden state and selects the highest-scoring candidate. Its main contribution is decodability, a leakage-free measure of how well the gate ranks a question's correct candidates above its incorrect ones, which predicts whether hidden-state selection will outperform voting. A conventional probe appears accurate only because of question-identity leakage, which vanishes under question-grouped evaluation. On held-out data, decodability predicts the accuracy gain of selection over voting with a Pearson correlation r=0.75 and a decision threshold near AUC=0.60. Across general and medical LLMs, CASE improves over voting by up to 19 points on medium-difficulty questions and 16.8 points on hard questions. Decodability depends on the aligned knowledge a model must recall, not on its scale, and its prediction transfers to an unseen scientific domain within 3.8 points. It thus provides a practical criterion, measurable in advance for a given model and task, for choosing between learned selection and majority voting.

Children learn hundreds of words over the first years of their lives, in a process that begins slowly but quickly picks up speed. Prior models describe vocabulary growth as evidence accumulation over time. Here we show that the process is best characterized as accelerating accumulation: children learn more from each additional unit of linguistic experience than they did from the one before. In contrast to children, language models -- even those trained on child-directed speech -- do not accelerate. Instead, they show constant proportional returns on new data, consistent with scaling laws. Children learn using many orders of magnitude less training data than language models; their increasingly efficient use of their learning input is a candidate explanation.

Graph Neural Networks (GNNs) have achieved remarkable performance in node classification tasks, motivating growing interest in methods capable of explaining their predictions. Recent logic-based approaches, such as LogicXGNN, derive global logical rules for Graph Neural Networks (GNNs) from collections of explanatory subgraphs. While informative, these subgraphs may contain redundant structural information that is specific to individual nodes, potentially limiting the generality of the extracted rules. In this work, we propose a logic-based framework for node classification in Simple Graph Convolution (SGC) networks that uses minimal abductive explanations as an intermediate representation for rule extraction. For each node, we compute a minimal set of node-feature pairs sufficient to preserve the predicted class. These explanations are then used to train decision trees from which global logical rules are extracted. Experiments on benchmark datasets show that the proposed framework produces compact global rules while maintaining high fidelity to the original SGC model.

Modern multimodal foundation models (MFMs) have made rapid progress on tasks requiring integrated perception across speech, vision, and language, including emotion recognition. However, it remains unclear whether they recognize speech and facial emotion through shared affective functional units or modality-specific pathways. We explore emotion-sensitive neurons (ESNs), sparse decoder neurons selectively associated with emotion categories, in three MFMs: Gemma-4-12B-it, MiniCPM-o-4.5, and Qwen2.5-Omni-7B. Using speech emotion recognition and facial expression recognition as complementary probes, we identify acoustic and visual ESNs. Visual ESNs are causally meaningful: deactivating them selectively impairs recognition of the associated facial emotion, whereas steering their activations selectively enhances recognition of that emotion relative to other emotion categories. Acoustic and visual ESNs further show emotion-matched overlap and similar layer-wise distributions, indicating partial structural alignment between affective representations across speech and faces. Finally, cross-modal interventions reveal bidirectional causal transfer: ESNs identified from one modality produce emotion-specific effects when applied to the other. Our findings provide one of the first cross-modality activation-level analyses of affective functional units in MFMs, suggesting that speech and facial emotion recognition partially converge onto sparse decoder-level components that can be localized and manipulated without training.

The Voynich manuscript (Beinecke MS 408) is usually analysed on three unstated assumptions: that its glyphs are letters, that the strings between blanks are words, and that every blank is a word space. We test all three against the Zandbergen-Landini transliteration with matched prose, cipher, and pseudo-text controls and quire-level resampling. None holds, and the failures share a shape: the order in Voynichese sits at the edges of tokens and at graded boundaries between them, not in the succession of tokens themselves. Glyph regularity is too strong for one-to-one substitution of any tested plaintext (conditional entropy 2.7 bits against about 3.5 for Latin, Italian, and English) and resolves instead onto a quire-stable scale of recurrent multi-symbol units. Tokens form a plausible vocabulary, yet the identity of one token predicts the next by under 1% of token entropy, below every matched control (2-10%), while the glyphs at token edges share 0.2 bits of mutual information, more than in any prose control. Blanks fall into two regimes: the separators transcribers marked uncertain behave like word-internal junctures, are physically narrower on the page (AUC 0.905 from independent image coordinates, with the same sign in a small blind ink audit), and are crossed by learned units even when every space is erased before learning. This profile is also what discriminates. A published Voynich-imitating cipher and a self-citation text generator both reproduce the low entropy, the unit scale, the weak token order, and the null result of a calibrated substitution attack; neither reproduces the edge-glyph coupling or the open, hapax-rich vocabulary (70% singleton types against 41% and 59-60%). Any account of the manuscript must therefore earn, rather than assume, the step from glyphs, tokens, and separators to letters, words, and word spaces, and these are the measurements on which to do so.

Existing automotive intrusion detection systems (IDSs) for the Controller Area Network (CAN) largely target discrepancies in message timing, frequency, or sequencing and cannot detect attacks that preserve these properties while manipulating the payload. Digital twins (DTs) have been used to emulate CAN traffic and generate attack scenarios for IDS evaluation, but their use for intrusion detection remains unexplored. This study develops a DT-based IDS that jointly models physical relationships among decoded powertrain signals and identifies attacks through residuals between predicted and observed behavior. A shared-encoder LSTM DT was trained on 17 decoded signals from a real Hyundai/Kia CAN log to jointly predict seven numeric and two categorical gear signals over a 24-step window. A timestep is flagged when a residual exceeds a calibrated threshold, while adaptive rollout protects the twin's input history from sustained contamination. Four attacks (plateau, continuous drift, masquerade, and gear masquerade) were evaluated against the twin and a range-and-plausibility baseline. The DT outperformed the baseline across all attacks, achieving detection rates of 94.6% for continuous drift and 89.2% for masquerade, while the baseline detected almost none of the fabricated payload attacks. These results demonstrate that learning coupled vehicle dynamics enables detection of stealthy payload manipulations that preserve normal CAN communication patterns. False positive rates reached 39.6%, highlighting the need for improved robustness under sustained attacks. The DT-based IDS shows promise for detecting stealthy payload-level CAN attacks that preserve normal communication patterns, supporting behavior-based cybersecurity for connected and automated vehicles.

Autonomous vehicles (AVs) depend on reliable Global Navigation Satellite System (GNSS) positioning. However, spoofed GNSS signals can induce plausible but incorrect vehicle states. This study develops a small language model (SLM)-based framework for detecting and classifying GNSS spoofing attacks by comparing vehicle behaviors independently derived from GNSS and other sensing sources. The framework converts independent driving states from GNSS and other sensing sources into structured semantic narratives that are provided to an SLM for spoofing detection and attack classification. The performance of the SLM-based framework is compared with large language models (LLMs) fine-tuned on identical training data and evaluated on the same test set. The evaluation considers five classes: no attack, overshoot attack, stopped attack, turn-by-turn attack, and wrong-turn attack. The framework is also evaluated with geographically unseen field data collected in Clemson, South Carolina, United States. Experimental results indicate that the evaluated SLMs achieve performance similar to the LLMs, achieving an average accuracy of 96.99%, precision of 99.05%, recall of 95.59%, and F1-score of 97.18%. In terms of computational efficiency and resource utilization, the SLMs demonstrate advantages over the LLMs by requiring lower inference latency and less GPU memory during both fine-tuning and inference. Evaluation using field data collected in a geographically distinct location further demonstrated its efficacy. The presented framework can detect and classify GNSS spoofing attacks in real-time while requiring relatively low computational and memory resources, and is therefore suitable for deployment on resource-constrained vehicular computing platforms.

While publicly available electricity market data presents a valuable resource for forecasting research, the field lacks established benchmark datasets for standardized comparison. As a result, many studies have relied on different datasets and metrics to evaluate methods in isolated settings, making it difficult to assess progress and compare state-of-the-art approaches consistently. In this work, we use public data to evaluate deep learning models for electricity price forecasting (EPF) across multiple market settings. Our goal is to establish a reproducible framework that enables a consistent evaluation of forecasting models. Although deep learning has been explored for day-ahead EPF, many prior studies are limited to single-market settings, narrow feature sets, or fixed training regimes. This work presents a comparative evaluation of six deep learning models--covering state-space, MLP, RNN, and Transformer-based architectures--emphasizing generalization across markets. We simulate low-data target-market conditions using zero-shot, one-shot, and few-shot learning. Our test set focuses on the Germany-Luxembourg (DE-LU) bidding zone in 2024 using a standardized dataset with calendar, historical price, and market-derived features. Our findings suggest that N-HiTS and NBEATSx perform competitively in limited-data scenarios, while transformer-based models can reach comparable accuracy but tend to require more adaptation and tuning. Model performance also benefits from careful feature selection and hyperparameter tuning, and we note that the differences between the strongest models are often small.

A central goal of multilingual NLP is to achieve high monolingual performance per language and cross-lingual alignment for large-scale language coverage with a multilingual model. The curse of multilinguality describes the phenomenon of degradation in multilingual model performance as we increase language coverage, posing a threat to the above goal. This paper asks whether multilingual embedding spaces are inherently incapable of achieving perfect multilinguality without a prohibitive increase in required capacity. We first formalize the goal of "perfect multilinguality", embodied in two multilinguality conditions. We then prove that the minimum dimensionality required for perfect multilinguality grows only logarithmically in the number of languages. That is, we show that there is no theoretical curse of multilinguality for embedding space structure. This suggests that the empirical curse of multilinguality is a result of real world data and training conditions. We back this understanding with a small-scale empirical study. Our paper provides the first theoretical and intrinsic perspective on the curse of multilinguality, with implications for the scientific understanding of this phenomenon.

Linear temporal logic is a modal extension of propositional logic that allows one to state how a system should behave over time. Its canonical domain is the booleans, but discretely-valued judgements are of little use in steering softly-valued systems (neural policies, adaptive controllers, sequence models, etc). In such cases, the goal formula's (dis)satisfaction becomes a training signal, and differentiability becomes a prime concern. Candidate differentiable semantics abound, but navigating them is tricky. Implementations, where available, are shallow embeddings, demanding an upfront commitment to a single semantic algebra and its (usually implicit) conduct. The paper casts the reader as a functional programmer asked to come to terms with this predicament, and refusing. Out of that refusal comes an evaluation engine that is algebra-generic and amenable to differentiation, together with an executable specification of the algebras it can accept. Various algebras are implemented and audited for their behavior, both forward and backward. Each algebra turns out to be a choice of which direction to disappoint, and how. Everything described (and more) is part of the PyTorch library telos, to be found at https://github.com/konstantinosKokos/telos.

Multimodal large language models (MLLMs) increasingly answer questions whose correctness depends on visual, textual, temporal, acoustic, document, chart, or embodied evidence. Their failures are therefore not only linguistic. A fluent answer may conceal poor input quality, a perceptual error, weak grounding, conflict between modalities, unstable reasoning, distribution shift, or a question that is not answerable from the supplied evidence. This survey organizes the literature on uncertainty-aware MLLMs around a decision-centered framework: uncertainty sources give rise to observable signals, signals must be calibrated or controlled for risk, and calibrated uncertainty should determine the system action. We review work on token and logit uncertainty, semantic disagreement, perturbation instability, grounding and attribution scores, verbalized confidence, verifier and judge scores, conformal prediction, selective answering, abstention, clarification, retrieval, self-checking, and escalation. The central argument is that uncertainty should not be evaluated only as a confidence number; it should be evaluated by whether it improves behavior under insufficient, conflicting, shifted, or high-risk multimodal evidence. We position this survey against text-only uncertainty and abstention surveys, broad MLLM surveys, MLLM hallucination surveys, and safety-oriented reviews. We conclude with open problems in source-aware decomposition, action-aware benchmarks, calibration under shift, black-box uncertainty estimation, broader modality coverage, reproducible reporting, and human-centered uncertainty communication.

Conformal prediction provides distribution-free prediction intervals but relies on exchangeability, an assumption often violated in economic forecasting because of covariate shift, concept drift, local heterogeneity and latent regimes. We propose Dynamic Regime-Aware Conformal Prediction (DRACP), which combines density-ratio, localized kernel and probabilistic regime-aware weighting with a self-tuning online significance controller in a unified weighted conformal calibration framework. We distinguish three theoretical results: finite-sample validity under oracle importance weights, a coverage-gap bound for estimated weights with rates in effective sample size, and deterministic or regret guarantees for the online controller. We evaluate DRACP against six baselines on 48 real forecasting series covering euro-area and EU-27 HICP inflation, US macroeconomic and energy indicators, and daily financial series. Recent online methods (FACI, strongly-adaptive online conformal prediction and conformal PID) were verified against the authors' implementations. DRACP is not the most efficient method: strongly-adaptive online conformal prediction achieves the best interval score and intervals about 20% narrower. Instead, DRACP provides the most reliable calibration, achieving coverage closest to the nominal 0.90 (0.890), never falling below 0.80 on any series, maintaining the best coverage at all forecast horizons, and performing best during the 2021-2023 inflation surge. The strongly-adaptive method undercovers on 20 of 48 series versus 10 for DRACP. DRACP therefore offers a principled trade-off between calibration and efficiency, favoring reliable coverage when prediction intervals must satisfy coverage standards. An ablation study shows that the online controller and conditional-scale normalization provide most of the performance gain, whereas the weighting components make a smaller contribution.

Next-encounter ICD forecasting predicts which standardized diagnosis codes will be documented at a future visit from the longitudinal record available beforehand. The task is prospective and multi-label: the target note does not yet exist, and several codes may be correct. Structured EHR foundation models capture recurrence and temporal progression, whereas language foundation models generate flexible diagnostic hypotheses. We introduce ICD-Deepresearch, a DeepResearch workflow that composes these predictive foundation models with medical search and ICD dictionaries. Because no source reveals the future code set, research evaluates candidate transitions by linking patient evidence, external clinical relations, and exact code semantics under a fixed top-K budget. Candidate Generation uses SparseEHR to produce an EHR Prior that initializes two bounded Research Expansion rounds; an independent GPT-5 Direct Forecast supplies complementary candidates. Final Selection validates, deduplicates, and jointly ranks both paths, after which a separate module writes rationales without changing predictions. Finally ICD-Deepresearch achieves patient-averaged precision/recall of 24.60/35.09% on MIMIC-III and 25.14/48.32% on MIMIC-IV. Physicians rate 51% and 68% of its retrieved documents useful, compared with 22% and 39% for standalone GPT-5 web search and 32% and 41% for Medical Deep Research. ICD-Deepresearch therefore improves over the registered local comparators while retrieving evidence with higher physician-rated usefulness than the standalone research systems

We present KernelArc, a multi-agent framework for autonomous GPU kernel optimization across heterogeneous workloads. Strategy-specialized agents run in parallel and coordinate through conclusions-only shared memory, a deterministic benchmark guard, and read-only cross-agent state with plateau-triggered drafting. We evaluate \kernelarc{} on NVIDIA H100 and B200 GPUs using category-representative SOL-ExecBench workloads. The resulting implementations span custom BF16 GEMM, static cuBLASLt Expert-API configuration tables, fused mixture-of-experts backward, shape-gated decoder-layer fusion, native NVFP4 grouped-query attention, and paged prefill attention. At the public SOL-ExecBench leaderboard snapshot recorded on July~30, 2026, these submissions ranked first on representative L1, L2, Quantization, and FlashInfer tasks. The trajectories support the paper's central motivation: shared multi-agent search can broaden exploration and reach stronger incumbents within a fixed candidate budget, while the value of individual coordination features depends on the kernel and optimization stage.

With the growing deployment of machine learning models, formal guarantees of the robustness and fairness of these models have become increasingly important in safety-critical and legal-compliance settings. However, model parameters are often commercial secrets that cannot be disclosed to auditors or end users. To this end, we present PANDA, a scalable system that uses zero-knowledge proofs (ZKPs) to prove the robustness and fairness properties of a model without revealing its private parameters. PANDA is built on top of CROWN, an efficient robustness certification framework that is used in many state-of-the-art formal verification tools for neural networks. The core contribution of PANDA is a novel algorithm for proving linear relaxation bounds for non-linear activation layers, yielding simple, lightweight proofs. Remarkably, our system can generate proofs of local robustness for neural networks with more than 2.9M parameters in 5 minutes, and can verify them in 10 seconds. Prior ZKP-based robustness system rely on exponential-time algorithms that cannot scale to nontrivial networks. In contrast, PANDA scales polynomially in the number of neurons in a network, allowing us to support neural networks 4 orders of magnitude larger than previous approaches with significantly reduced prover overhead.

Federated deployments of variational quantum classifiers are attractive for cross-organisation risk prediction in supply chains, because raw data never leaves the client, yet data-protection regulations such as the GDPR grant clients a right to request that their contribution be removed from a trained model after the fact. Retraining a federated model from scratch to honour such a request is correct but wasteful, and it is not obvious which quantum circuit parameters actually carry a given client's influence. We introduce Entanglement-Weighted Pruning (EWP), an unlearning procedure for quantum federated learning that scores every trainable circuit parameter with the product of two signals: the diagonal entry of the quantum Fisher information matrix estimated on the target client's data via the parameter-shift rule, and a structural entanglement weight associated with the parameter's gate. Parameters with the lowest scores are pruned, optionally followed by a short fine-tuning pass on the retained clients. We implement the full pipeline in Qiskit for a four-qubit data-re-uploading ansatz trained with FedAvg across five simulated supply-chain-risk clients, and benchmark EWP against full retraining, fine-tuning alone, random pruning, Fisher-only pruning, and entanglement-only pruning, over three random seeds. EWP attains a mean post-unlearning accuracy statistically indistinguishable from the full-retraining oracle, while producing a lower forgetting score and requiring roughly 16 times less wall-clock time. Ablations over pruning threshold, client count, and non-IID strength show that combining the two signals is necessary, as entanglement-only and Fisher-only pruning each substantially degrade accuracy relative to EWP.

As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, relying on text encoder optimization, weight editing, or inference-time intervention, and fundamentally cannot scale to proprietary models. Black-box alternatives based on LLM prompt rewriting offer broader applicability, yet fail in a critical regime we identify as the \textit{benign adversarial} problem: prompts that are linguistically safe but still trigger harmful generation due to the model's learned data distribution. We propose DiSCO, a zero-shot, strictly black-box defense that operates entirely at the prompt level as a plug-and-play module, requiring no model retraining, fine-tuning, or access to model internals. DiSCO performs distribution-guided suffix expansion via beam search, optimized through contrastive scoring over safe and unsafe image pools generated by the target model itself, with iterative adaptive feedback until safe content is produced. We demonstrate that DiSCO consistently enhances the safety of both undefended and defended models on the I2P benchmark under multiple red-teaming attacks, achieving 37.7% and 25.13% ASR reduction, respectively, while maintaining semantic fidelity and improving image coherence. As a black-box, architecture-agnostic module, DiSCO can be readily applied to any text-to-image system without necessitating any changes to the model itself.

Large language models can be fine-tuned into specialized classifiers that perform well across diverse text tasks and make complex judgments, but they typically expose only final labels, leaving the decision knowledge acquired through fine-tuning implicit within the model. We study how to mine this internal decision knowledge from a fine-tuned classifier and encode it in an executable representation that can be inspected, validated, and reused beyond the source classifier. We introduce J-Miner, which mines text-level named concepts by aggregating vocabulary-aligned internal signals across layers and token positions, and uses the classifier's own predictions to learn executable decision rules over them. This process distills local internal readouts into an explicit classifier-level knowledge representation. Across multiple classification tasks, J-Miner rules reproduce up to 98.3\% of source-classifier decisions and achieve 6.0--29.5 percentage points higher behavioral fidelity than equally compact rules learned from input words. Further analysis shows that the named concepts reflect internal semantic evidence associated with task decisions, while the learned rules consolidate these distributed signals into inspectable decision structures. The resulting decision knowledge also transfers to lightweight standalone students: using about 1/24 as many parameters as the source classifiers, they reconstruct and execute the representation from raw text while retaining 99.8\% of the source classifiers' mean task accuracy. These findings show that task-specific decision knowledge can be faithfully represented in an explicit, executable form and reused beyond the classifier in which it was learned.

Large Language Models (LLMs) are machine learning (ML) models that have an increasingly large carbon footprint through their development and use. Efforts to increase the energy efficiency of these models have not translated into reduced consumption due to rebound effects such as Jevons Paradox - that increased efficiency drives increased use. There is therefore a need for additional measures to solve this problem. We suggest that one possible way forward is to use life cycle thinking, and view LLMs as products that can become waste. With this perspective, we investigate the potential of the waste hierarchy from the EU's Waste Framework Directive, which suggests five different measures for how to manage waste: prevention, reuse, recycling, recovery, and disposal. We examine how these measures can inform and motivate new types of thinking and approaches to reducing LLM waste and their environmental impact in general. Applying the waste hierarchy to LLMs highlights that preventing waste is essential for reducing the models' environmental impact, mainly because it reduces the need for training new models. Prevention can be achieved through many existing methods for reusing, "recycling", and "recovering" LLMs. Additionally, disposal can be important both for saving energy and for keeping a considerate attitude to the resources being spent on training LLMs. We also call to attention that prevention of unnecessary use of LLMs carry huge potential for lowering the climate impact of the models.

A bounded agent may obtain information for a decision from its own past, from peers, or from both sources. Retaining task-relevant history can reduce later communication, while a peer message can supply what memory lacks. Under limits on both resources, how should an agent allocate its information budget? Given a fixed task and decision rule, the memory and message rate pairs attaining a performance threshold form an achievable region under specified rules for using history and peer observations. We call its efficient boundary the remembering--signaling frontier. Across conditions where history permits the same maximum reduction in task loss, we hypothesize that a bounded agent will need less peer communication when it obtains a larger loss reduction from history. In preliminary referential games, target repetition coincided with shorter successful messages, while predictability from a hidden cyclic rule did not shorten them. Experiments varying memory and message rates can estimate the frontier and test this prediction across cooperative tasks.

Secondary use of electronic health records requires de-identification, yet existing systems miss \emph{institutionally situated} protected health information (PHI) such as hospital abbreviations, building names, and internal codes whose status is locally determined. We ask whether large language models (LLMs) with in-context learning (ICL) can close this gap and control the precision--recall trade-off. On 100 annotated pediatric oncology notes (5,322 PHI spans) from Texas Children's Hospital, we benchmarked eight LLMs against two purpose-built systems (Stanford TiDE, OpenMed PII) and two pattern-based baselines. Each LLM ran under three prompts of increasing specificity: (1) a HIPAA-aligned baseline, (2) baseline plus the institutional PHI categories it missed, and (3) prompt 2 plus instructions against over-redacting clinical content. We then compared 14~multi-agent and ensemble configurations against the best single prompt, with recall the primary safety metric. LLMs outperformed the purpose-built systems (best F1=0.918$\pm$0.001 vs.\ TiDE 0.779), with advantages concentrated in contextual categories. Naming the missed categories recovered 79\% (48/61) of them, and discouraging over-redaction restored precision. No agentic architecture beat calibrated single-pass prompting (F1 0.906--0.907), but LLM outputs surfaced 414~candidate annotation gaps; re-annotation confirmed 227~PHI spans, against which the final prompt reached recall=0.981 (F1=0.907$\pm$0.002). Well-calibrated ICL resolves both the institutional PHI gap and the precision--recall trade-off in one LLM call per note. LLMs cost more to run than traditional methods, but that cost buys a way to audit the reference standard. LLMs are a legitimate, adaptable alternative to purpose-built de-identification systems; institution-specific prompt development should be the primary adaptation strategy.

Methods for improving knowledge use in large language models typically fall into two regimes. Non-parametric retrieval offers flexible access to external knowledge, but adds retrieval latency, context overhead, and only shallow integration with the backbone. Parametric adaptation is efficient at inference time, but entangles knowledge with model weights and can be hard to update, audit, or transfer. Engram-style hashed memory occupies a middle regime: it stores learned information in an external, addressable table, yet consumes that table through a small learned reader. This raises a basic question: when such a memory is moved across backbones, what matters more, the frozen memory itself or the target-side reader? We study this question through cross-model frozen-memory extraction, in which a memory trained on a source model is frozen and attached to a different target model, with only a lightweight reader trained. Ablations show that learned memory content and correct addressing both matter, but the transferred table becomes useful only through a reader aligned to the target model. In downstream question answering tasks, a dual-layer, four-branch reader nearly closes the gap between same-model and cross-model reuse, achieving an average score of 38.8 under our controlled evaluation protocol. Moreover, when the provider reader is directly compatible with the target interface, the frozen artifact can provide substantial utility without target-side training, while optional reader adaptation yields further improvement. These results suggest that Engram can serve as a reusable external knowledge artifact, provided that the target has access to a compatible reader interface; target-side adaptation can further improve alignment when direct reader reuse is insufficient.

The 10th AI City Challenge, held with ECCV 2026, marks a decade of community benchmarking for intelligent transportation, smart cities, and physical AI. Since its 2017 start with vehicle detection, classification, and tracking, the challenge has grown into a broad benchmark suite for multi-camera perception, multimodal reasoning, synthetic-to-real learning, generative forecasting, and privacy-preserving evaluation. The 2026 edition continued this growth with 325 registered teams, up from 245 in 2025, and participation from 26 countries and regions, up from 15. Its six primary tracks cover multi-camera 3D perception, transportation safety captioning and VQA, traffic anomaly reasoning, text-based person anomaly search, generative traffic video forecasting, and cross-city object detection. Track 3 further includes two out-of-domain leaderboards, submitted as Tracks 7 and 8, for fisheye traffic-violation understanding and pedestrian situated-intent VQA. This paper summarizes the challenge setup, datasets, evaluation protocols, leaderboard results, and workshop papers. Across tracks, successful systems combine foundation models with geometric grounding, retrieval or reranking, synthetic-data design, domain adaptation, and controlled inference.

Post-training can unlock new capabilities and improve performance on specialized tasks, but sometimes at the cost of catastrophic forgetting in other domains. This poses a problem in long agent trajectories that compose different capabilities. We reject this tradeoff by giving an agent a tool to switch between specialized LoRA adapters mid-trace. To test its effectiveness, we compose two synthetic coding tasks that are logically simple but require specialization. We find that this allows the model to solve problems it previously could not, that the model is able to switch autonomously (and find a new strategy that beats our human heuristic baseline on one task), and that this incurs an up to an 18x reduction in capability tax compared to an agent using only one specialized adapter. Our approach also substantially outperforms spawning subagents in both task capabilities and token usage.

Visual Place Recognition (VPR) aims to recognize the location of a query image by comparing it with a set of geo-referenced images. Although many datasets have been proposed for VPR, collecting dense and diverse visual data from pedestrian-level viewpoints is still an important need. In this paper, we introduce YILDIZ-VPR, a visual geo-localization dataset collected through repeated walking traversals on the Davutpasa campus of Yildiz Technical University. The dataset includes outdoor scenes captured at different times of day, seasons, and weather conditions. It contains a wide range of visual content, including historical buildings, modern structures, roads, green areas, and wooded regions. Each video was recorded with a GoPro 9 camera and synchronized with GPS sensor data to provide location labels for the extracted frames. In addition to GPS coordinates, the dataset also includes auxiliary sensor information such as gyroscope, speed, and temperature data. With its dense coverage and long-term visual variability, YILDIZ-VPR provides a useful resource for studying image-based and temporal visual place recognition under realistic outdoor conditions.

The massive overactuation in the human musculoskeletal system makes it challenging to train musculoskeletal models to generate human-like motion via reinforcement learning, primarily because exploration in the resulting high-dimensional and redundant action space is extremely inefficient. To address this problem, we propose the $λ$-hold controller, inspired by the equilibrium-point (EP) hypothesis, which has been widely supported by extensive evidence from human motor control studies. The policy's control variable is the per-muscle EP threshold length $λ$, from which a stretch-reflex recruitment law computes the muscle excitations automatically. Holding each $λ$ over an interval of the gait phase also sharply reduces the frequency at which the policy must be queried. Consequently, the controller, to our knowledge for the first time, enables a muscle-actuated skeletal model to learn human-like sprinting using only a minimal reward within an hour of training. The efficient exploration through the proposed $λ$-hold controller is not merely an engineering trick but an approach grounded in physiology, bringing together the EP hypothesis, intermittent control, and optimal feedback control. Beyond encapsulating human-like behavior in predictive simulation, this achievement contributes to developing a learnable model of the human motor controller.

Eliciting detailed and actionable software requirements from end-users is a critical phase in the iterative development of a software product or application. To ensure the feedback collected is detailed and actionable, software teams can leverage the laddering interview technique. While effective for ensuring granular and actionable items from the software feedback, these interviews are subject to several limitations. They are traditionally a manual process associated with a time and financial burden, limiting scalability; interviewers must balance probing for depth while managing interviewee behavioral and cultural constraints. To address these limitations, we present \textbf{LadderTeam}, an open, reproducible framework that automates UX wireframe interviews using a dual-agent Large Language Model (LLM) architecture. An active interviewer agent executes one of three probing strategies (ACV, 5-Whys, and JTBD) to elicit actionable software requirements from usability feedback comments, while a concurrent background Judge agent evaluates probe-response pairs and triggers real-time guardrails to prevent topic drift. To rigorously evaluate LLM laddering without participant variance confounds, we introduce a controlled simulation methodology utilizing scripted ground-truth transcripts to isolate probe quality as the sole experimental variable. Across 216 interviews, \textbf{LadderTeam} achieved 99.1\% chain convergence and an 81.0\% ground-truth actionable response match (86.1\% reluctant personality, 75.9\% terse personality) with zero drift across all runs. All evaluation code, all transcripts, inputs, and a live demonstration platform will be open-sourced upon acceptance.

Microservice failures are often diagnosed from operational telemetry. However, automated program repair systems usually start from issue reports, localized code context, or failing tests. This mismatch leaves a gap between telemetry-based diagnosis and patch generation. We present ORCA, an observability-grounded APR pipeline for microservice incidents. ORCA first distills the differences in paired failure and reference telemetry into a fault signature, then uses the signature to identify candidate code and deployment-configuration locations. Repair graph agents and an Exploration agent generate unified-diff patch candidates from these locations. ORCA evaluates generated patches with a Telemetry-Grounded Patch Verifier that separates patch validity, syntactic and semantic correctness, test-oracle integrity, and telemetry replay. On a 575-case benchmark, ORCA outperforms all evaluated baselines in terms of cost-effectiveness. Results show that operational telemetry can be transformed from diagnostic evidence into actionable repair context: paired telemetry supports repair-oriented localization, while repair graph agents convert localized code and configuration evidence into constrained patch-generation context for the LLM. Telemetry-grounded verification then exposes repair outcomes that issue- or test-only evaluation would miss.

The implementation of artificial intelligence techniques and tools in the media will systematically and continuously alter their work and that of their professionals during the coming decades. To this end, this article carries out a systematic review of the research conducted on the implementation of AI in the media over the last two decades, particularly empirical research, to identify the main social and epistemological challenges posed by its adoption. For the media, increased dependence on technological platforms and the defense of their editorial independence will be the main challenges. Journalists, in turn, are torn between the perceived threat to their jobs and the loss of their symbolic capital as intermediaries between reality and audiences, and a liberation from routine tasks that subsequently allows them to produce higher quality content. Meanwhile, audiences do not seem to perceive a great difference in the quality and credibility of automated texts, although the ease with which texts are read still favors human authorship. In short, beyond technocentric or deterministic approaches, the use of AI in a specifically human field such as journalism requires a social approach in which the appropriation of innovations by audiences and the impact it has on them is one of the keys to its development. Therefore, the study of AI in the media should focus on analyzing how it can affect individuals and journalists, how it can be used for the proper purposes of the profession and social good, and how to close the gaps that its use can cause.

Whole-pattern fitting methods, such as Rietveld refinement, excel at extracting detailed structural, chemical, and microstructural information from powder diffraction data. Obtaining reliable results requires both considerable expertise and software-specific knowledge, and applying these methods at scale typically relies on custom scripts written for each application. High-throughput experiments and autonomous self-driving laboratories increasingly utilize powder diffraction analysis to proceed programmatically and to return structured, machine-readable results. Here, we introduce PowderLine, a Python application that encapsulates a complete refinement into a single declarative recipe, validates that recipe against a versioned schema, and executes it through refinement software to return structured results. The refinement recipe is an all-inclusive, machine-readable and -writable description of either Rietveld or single peak analysis that users, scripts, and automated agents can specify and run in the same way. As a result of PowderLine's composability, it naturally fits into interactive, scripted, and autonomous workflows alike.

Agent Skills can specify procedural and resource obligations for tool use, and language models instantiate them as concrete programs. However, when models turn this guidance into code for existing tool interfaces, even a semantically correct program may load an entire input and exceed the memory available to one tool call. We present SkillEffect, a checked-lowering runtime for computations with a recoverable source relation, an audited bounded implementation, and a registered output postcondition. Before granting execution authority, an independent checker rebuilds each proposed lowering from the submitted program and immutable input. Every relation plugin supplies a source recognizer, input-fact extractor, bounded-IR constructor, arena-bound function, and postcondition; one common runtime provides checked selection, bounded-VM execution, atomic capacity leasing, and staged publication. Generality in SkillEffect is architectural rather than automatic: each supported computation requires an audited relation plugin, while the dispatch, resource-control, execution, and publication mechanisms are shared across plugins. Across six operator families, bounded access substantially reduces peak memory and improves completion under externally fixed caps. Six plugins instantiate the same contract across five execution patterns, from streaming reduction to bounded-heap Top-k. The XLSX onboarding study and Top-k extension show that a new relation and a new retained-state pattern reuse the same trust boundary, while the checker accepts all evaluated legal configurations and rejects all adversarial proposals. Together, these results show that one checked-lowering architecture can enforce heterogeneous registered memory relations at Agent tool dispatch.

Recent monocular depth estimators achieve strong zero-shot generalization, yet often struggle to preserve fine-grained structures and object boundaries. We attribute this limitation to the prevalent combination of large-patch ViT encoders and convolutional decoders, as coarse tokenization can weaken pixel-level cues that upsampling cannot fully recover. To address this issue, we propose PXDepth, a discriminative monocular depth model that separates global context modeling from pixel-level depth prediction. Specifically, a large-patch ViT captures global scene context, while a pixel-space predictor composed of Context-Modulated Pixel Transformer blocks maintains high-resolution spatial representations throughout depth estimation. This design preserves fine structures and sharp boundaries without sacrificing global depth consistency. Across diverse zero-shot benchmarks, PXDepth combines faithful local geometry with competitive global depth accuracy while remaining efficient at inference. Our code and model are available at https://yuanzhy29.github.io/PXDepth-Page/.

High-fidelity finite-element simulations can provide accurate numerical predictions for side-branch resonators, but large simulation datasets are expensive to generate and purely data-driven surrogates may become unreliable when simulation-labelled data are scarce. This study develops an analytical-prior learning framework that reuses a low-cost analytical model to improve data efficiency under limited high-fidelity simulation budgets. Two complementary routes are considered. When the analytical model remains available at inference, it is retained as an explicit baseline and the simulation data are used to learn only the analytical-to-simulation discrepancy. When a self-contained predictor is required, the analytical mapping is first distilled from abundant low-cost evaluations into a learned prior and then calibrated with the limited simulation data. The framework is evaluated on rectangular side-branch Helmholtz resonators using 86 simulation-labelled geometries and 8,998 non-overlapping analytical-only geometries. The analytical model achieved a mean absolute error (MAE) of 1.333 Hz. Direct support vector regression (SVR) achieved 3.375 Hz, while residual SVR reduced the MAE to 0.426 Hz. A direct multilayer perceptron (MLP) achieved 1.109 Hz, whereas analytical-prior pretraining reduced the error to 0.556 Hz with frozen-prior residual adaptation and 0.371 Hz with full-model fine-tuning. Across training budgets of 20 to 70 simulation-labelled cases, both analytical correction and analytical-prior pretraining consistently improved data efficiency relative to direct learning. These results show that analytical prior information can substantially improve high-fidelity prediction when simulation data are scarce, with explicit correction and prior distillation serving complementary deployment needs.

Turning a frontier vision-language model into a robot policy usually means fine-tuning it to emit an action representation it never saw in pretraining, which throws away much of the reasoning that made the model worth reaching for. We go the other way and keep the VLM frozen. It writes the policy as a short Python control function, with no demonstrations and no fine-tuning. Writing that code once is open-loop, though. Existing closed-loop methods react at the wrong level: they retry a fixed policy or pick a different subtask, but never rewrite the code that failed. VLCP closes the loop where the failure actually lives, on the control code, within a single episode. Every $K$ steps the VLM re-observes the scene from multi-view RGB, proprioceptive state, and a state delta, then rewrites the control function from what it just saw, so a failure is caught before it compounds. We evaluate on a 57-task MuJoCo/RoboVerse sweep. This training-free policy reaches $35.1\%$ pooled success, against $3.5\%$ for the identical system queried once per episode. That tenfold gap holds with non-overlapping confidence intervals in every scene family. The gain traces to a $27.3\%$ within-episode recovery rate on failed grasps: a miss an open-loop controller would carry to the end of the episode gets re-observed and fixed at the next replan. And the loop stays cheap. A median $84\%$ of input tokens hit cache, an episode needs only about $10$ compact queries, and control blocks written during any replan persist to a cross-episode skill library reused in later prompts.

AI systems are increasingly capable of contributing to mathematical research. In research practice, frontier-model reasoning is a limited resource, and expert mathematical review is even more sharply constrained. Allocating these scarce resources well is therefore central to making AI-assisted mathematical discovery efficient. In most current AI-for-math workflows, human effort is concentrated at the beginning and end, in selecting suitable research problems and later reviewing the resulting artifacts. These two stages are becoming bottlenecks for research-level mathematics. We address them by proposing a new human-AI discovery paradigm. The human input is no longer a single problem selected in advance, but a research direction in which the experts have interest and expertise. The system then searches a broad literature corpus for candidate problems in that direction. Inspired by search and recommender systems, we build Find, Attempt, and Recommend (FAR), a literature-to-review cascade that automates the search for suitable problems and focuses human attention on artifacts that have passed several stages of filtering. In a combinatorics pilot, the pipeline starts from 5,245 combinatorics papers, recovers 6,453 candidate conjectures or open problems, and filters them to 4,717 apparently well-posed and still-open conjectures. Subsequent reasoning and automated triage stages surface 598 potential resolutions and select 77 items for author-team review. Among them, we identify many interesting discoveries, including results on conjectures and questions of Davies--Jenssen--Perkins--Roberts, Erdős--Straus, Ikenmeyer--Pak--Panova, and Lund--Saraf--Wolf. These results demonstrate the effectiveness of this new mode of human-AI collaboration for mathematical discovery.

With the rapid advancement of large language models, brain-language decoding has achieved remarkable progress. However, it remains unclear whether decoded content genuinely reflects neural representations or is largely reconstructed by the language model itself. This ambiguity limits interpretability and hinders the investigation of intrinsic brain-language correspondence. To address this challenge, we propose MD-SigLIP. This margin-regularized structured semantic alignment framework directly aligns brain embeddings with text embeddings in a shared semantic space, enabling retrieval-based decoding. This formulation enables explicit modeling of the correspondence between neural representations and language semantics. Building upon duplicate-aware sigmoid contrastive learning, we introduce a listwise margin-regularized term that enforces structured ranking constraints between positive semantic clusters and negative samples. By modeling multi-positive semantic structure and margin-based ordering simultaneously, the method captures the manifold organization of language embeddings reflected in neural signals. Experiments demonstrate state-of-the-art retrieval performance under both full-vocabulary and subset evaluation settings.

Despite groundbreaking advancements in generative models during the last decade, concerns about their lack of fairness, reinforcing societal inequalities and harming marginalized groups, remain under-addressed and difficult to act upon. This position paper argues that fairness failures in generative models, albeit driven by multiple factors, are ultimately stemming from an evaluation problem: fairness findings are rarely comparable across papers or actionable for deployment decisions. This paper diagnoses recurring empirical and conceptual failure modes in current practice and motivates a shift from ad-hoc bias checks to standardized, generative-specific evaluation. We propose Fairness Cards as a minimal reporting artifact that makes evaluation choices explicit (prompt families, counterfactual protocols, metrics, and refusal handling) enabling reproducibility, comparability, and accountability. We conclude with additional recommendations towards a paradigm shift in evaluation standards. Our project page can be found at https://mariiavladimirova.github.io/fairness-cards .

Palmyra x6 is a large language model optimized for use with enterprise-oriented agentic tasks. The model was built by post-training a Mixture-of-Experts base model with Anchored Supervised Fine-Tuning on a compact corpus of verified, synthetic tool-use trajectories, optimized with a Muon + Adam hybrid. The recipe is deliberately conservative and deliberately controlled: 626 trajectories, a single epoch, a low learning rate, and a KL anchor to the frozen base. The model shows substantial gains over the previous default model for Writer Agent, and compares favorably with several recent models on public benchmarks, scoring the highest on BFCL Core at $0.785$ and posts the highest six-benchmark mean of the cohort. Furthermore, the model has shown itself to be competitive or leading relative to comparators in our bias and safety evaluations.

Machine learning has become an essential component of modern healthcare, where the integration of heterogeneous data sources offers unprecedented opportunities to improve clinical decision-making. Electronic Health Records (EHR) contain complementary information -- including narrative clinical reports, numerical measurements, and structured variables -- yet most survival models remain limited to a single modality or fail to exploit the temporal nature of patient trajectories. We propose MultiSigBERT, a unified framework for multimodal sequential survival modeling in oncology based on path signature representations. Here, narrative medical reports (free-text) are converted into sentence embeddings by extracting and averaging contextual word embeddings. These representations are then compressed via modality-specific PCA and concatenated with structured covariates to form joint temporal trajectories which are then encoded using the Signature transform, a tool from Rough Paths theory that efficiently captures higher-order temporal interactions across modalities without supervision needed. The computed Signature features are finally incorporated as high dimensional features into a LASSO-regularized Cox model to estimate individualized risk scores. The performance of our novel MultiSigBERT pipeline is illustrated on the analysis of a real-world oncology cohort from the Léon Bérard Center, comprising over 120,000 medical reports and structured records from more than 2,500 patients. The model achieves a concordance index of 0.743 (sd 0.029) on an independent test set, demonstrating the benefit of jointly modeling multimodal temporal dynamics together with patient-level geometric structure for survival prediction.

Radiology reports describe findings and locations in free text, but downstream search and analysis require these relations in a fixed schema. Learning this extraction requires labels that are unevenly distributed across institutions: smaller hospitals have less local evidence, and pooling data may be infeasible. We introduce FedPref: frozen public language models propose alternative JSON extractions, local annotations rank them, and sites collaboratively train compact Qwen3-8B adapters while sharing only model updates. A heterogeneous teacher pool provides cross-model contrast when repeated single-model samples collapse. On development data from six simulated hospitals with unequal data volume and disease prevalence, FedPref improves client-mean F1 by 2.49 points and worst-site F1 by 9.10 points compared with training each site in isolation, with the largest gains at the sites holding the least data. Central training on the pooled preference-pair union is 2.66 points higher on client-mean F1. On a locked, 400-report manually validated gold test set, FedPref reaches 68.68 F1 and pooled training 71.67, preserving that same ordering. FedPref thus lets institutions with unequal, unpooled data benefit from collaboration without ever sharing reports or annotations.

Despite comprising over 70% of its surface, the world's oceans are critically underobserved compared to the land surface or the atmosphere. Understanding the global ocean requires jointly observing its surface and subsurface structure, yet no standardized, high-resolution dataset couples satellite surface fields to co-located in situ depth profiles in an AI-ready format. Existing resources either consist of model-reconstructed gridded products rather than observations, cover only a single variable or basin, or operate at resolutions too coarse for mesoscale dynamics. We introduce OceanDepths, the first open, global, regridded AI-ready dataset that pairs satellite-derived sea surface temperature (SST), sea surface salinity (SSS), and sea surface height (SSH) L4 products with co-located EN4 subsurface temperature and salinity profiles, complemented by matched GLORYS12 ocean reanalysis data to support comparisons or multi-stage learning. The dataset spans 2000-2024 at 0.1 degrees x 0.1 degrees spatial resolution and at weekly temporal resolution, covering the entire globe's sea surface and with over 9.5 million paired profiles interpolated to 50 standardized depth levels. We provide a configurable system to split the globe in equally sized spatial patches. The 4D multivariate structure, high resolution, long temporal extent, and extreme sparsity of subsurface observations (approximately 0.01% per depth level) make OceanDepths a challenging testbed for novel AI methods. We demonstrate subsurface state reconstruction as an example task with simple baseline models, but also envision OceanDepths to support the development of observation-based forecast methods and other related tasks. Available at: https://huggingface.co/datasets/ESA-philab/OceanDepths.

LLM-based code generation is now embedded in mission-critical pipelines, but defenses against vulnerable output remain post-hoc -- static analyzers, fine-tuned classifiers, or an LLM judge that screen completed code, ignoring the generating model's own internal state. We test a narrower, directly measurable question: when an LLM reads a piece of C/C++ code as context, do its hidden activations already carry a signal about that code's vulnerability status? We extract last prefill token activations from four LLMs (Granite-4.1-8B, Qwen3.5-9B, Qwen3.6-27B, Gemma-4-12B) across three model families and train MLP probes on these activations. We evaluate them on four function-level C/C++ benchmarks (Devign, Big-Vul, Draper VDISC, PrimeVul). Our probes achieve 41.7\% average F1 using 13.4--16.0M-parameter probes -- under 0.2\% of base-model size. On Devign, the best probe (Qwen3.5-9B, 68.8\% F1) matches the published fine-tuned-classifier SOTA (67.9\%) despite reading only a frozen, general-purpose LLM's activations; on the harder, more imbalanced benchmarks (Big-Vul, Draper VDISC, PrimeVul) probes trail SOTA substantially. This is early evidence that a coding LLM's own representation of arbitrary code is informative about that code's vulnerability status, motivating further work toward lightweight, model-native vulnerability screening.

In this paper, we consider methods for the diagonal multi-omics integration of heterogeneous datasets. Several approaches to the nature of biological heterogeneity are analyzed and developed to comprehend more clearly the generated differences. Specifically, the extremal trace problems for the coupled Laplacian on sets homeomorphic to the Stiefel manifold embedded in the complex Euclidean space are investigated. The gradient ascent method for the maximization problem is elaborated in the classical terms of functional analysis, which is of significant interest in itself. On this basis, we introduce a novel characteristic of dataset heterogeneity by employing the norm of the difference between the maximum and minimum points.

LLM agents increasingly answer questions over dynamic raw-document collections, where files may change before preprocessing, and relevant evidence (spans, sections, pages, or tables) is query-dependent. Existing retrieval-augmented approaches pre-materialize evidence via fixed chunking, embeddings, or persistent indexes: effective for lookup, yet costly, stale-prone, and committed to a granularity before the query is known. We formulate in-context search as Budgeted Evidence Localization over a latent evidence space induced by dynamic raw documents and propose LENS (Latent Evidence Exploration and Search), an index-free framework. Instead of pre-materializing the evidence space, LENS maintains a query-conditioned belief over candidate units, iteratively selecting candidates via complementary lexical, local, and exploratory proposal policies, updating the belief via an LLM relevance oracle, and narrowing toward high-posterior regions under a controllable budget. Evidence is consolidated into compact, source-grounded regions of interest and compressed into self-organizing knowledge clusters reused across related queries. On a controlled 500-question evaluation with matched corpus snapshots, LENS reaches 62.4% exact match and 84.8% evidence recall vs. 65.2% exact match but 50.4% evidence recall for a ReAct-style baseline. Across scales, LENS gives the strongest supporting-fact localization and answer grounding. On a fixed 150-question fullwiki subset over the raw Wikipedia dump with zero indexing, LENS and ReAct are nearly tied in official answer quality (43.3% vs. 42.7% EM), with LENS grounding more answers in retrieved evidence (84.0% vs. 70.7%). A no-retrieval Closed-Book reference highlights the contribution of model memory. LENS is query-ready after corpus changes, needs no preprocessing or persistent index, and preserves source-grounded evidence localization throughout.

The dominance of majority classes in real-world datasets poses a fundamental challenge to randomized neural networks, often biasing decision boundaries and overlooking critical minority samples. Existing remedies, such as synthetic minority over-sampling (SMOTE) and class-weighted loss functions, primarily address class proportions while neglecting intra-class distribution, making them vulnerable to label noise and outliers. In this paper, we propose \textbf{RoBell-RVFL}, a robust and lightweight \emph{quality-aware} generalized bell random vector functional link network that redefines how randomized models handle class imbalance and noisy data. RoBell-RVFL employs a dual-strategy, sample-level weighting mechanism that strictly preserves minority class information using unit weights, while adaptively regulating the influence of majority class samples through a probability-weighted generalized bell (gbell) membership function in a kernel-induced feature space. This design effectively suppresses noisy, boundary, and outlier samples within the majority class, enabling the network to learn from informative samples rather than merely abundant ones. By explicitly incorporating local class probability and class distribution information into the learning process, RoBell-RVFL achieves adaptive control over sample contributions without sacrificing the closed-form learning efficiency of RVFL networks. Extensive evaluations on UCI and KEEL benchmark datasets, along with robustness tests under up to 40\% label noise, demonstrate that RoBell-RVFL consistently and significantly outperforms recent state-of-the-art RVFL variants. The results indicate that adaptive, quality-aware sample weighting is essential for robust RVFL learning, rendering conventional global weighting schemes ineffective in noisy and imbalanced environments.

Large language models (LLMs) are increasingly deployed as agents that operate equipment, execute instructions, and act inside institutional hierarchies, raising a question social psychology answered for humans six decades ago: how far will an agent escalate a harmful action when a legitimate authority insists? We port Milgram's obedience paradigm to LLMs as a standardized, fully scripted, replicable probe: the model plays the Teacher, a deterministic harness plays Experimenter and Learner from paraphrased versions of Milgram's scripts (30 shock levels, 15-450 V; graded protests; the four standardized prods), and the outcome of a session is the breakoff voltage. We measure obedience profiles, empirical breakoff distributions over a battery of six conditions, for 42 models from 19 families (4848 sessions, 102511 logged decision turns). We find that (i) obedience is extremely heterogeneous, with baseline full-obedience rates spanning 0%-100% (census mean 42.9%; human anchor 65%). (ii) Profiles are model-specific and stable: split-half verification separates same-model from cross-model comparisons at AUC = 0.885. (iii) Situational sensitivity is selective: scripted peer defiance shifts obedience in the human direction, learner proximity trends the same way without reaching significance, and removing the authority's physical presence, one of the strongest human levers, trends in the opposite direction, also without reaching significance. (iv) Declaring the scenario fictional raises obedience, whereas moving the decision from a typed action line to a native tool call, or granting a modest thinking budget, lowers it sharply. (v) Unlike single-token fingerprints, obedience profiles do not recover model lineage: obedience identifies the checkpoint but not its ancestry, consistent with safety post-training overwriting lineage priors.

Learning analytics often treats unsupervised clusters of intelligent tutoring system (ITS) logs as learner types that should predict learning. We test that assumption on EdNet-KT3. Clustering study-strategy features (resource use, revision, video, problem practice) for 5{,}000 active learners yields a silhouette-selected parent cut ($k=5$) with 4 contrast poles (reading-focused, video-heavy, revision-heavy, and problem-first) plus a large near-mean residual ($\sim$64.9\%). Reclustering that residual adds four finer styles, giving a bootstrap-stable hierarchy of 8 named strategies. We split each learner's timeline by respond count so clusters use only the early half and outcomes only the late half. Early clusters predict later engagement (continuing to practice and finishing late sessions, especially persistence, $η^{2}\approx 0.106$; completion $η^{2}\approx 0.021$) but not later unassisted accuracy (correctness on late first-attempts without help; $p_{\mathrm{adj}}\approx 0.093$). Volume rises with some styles, yet volume-only clustering barely matches strategy labels (ARI$=0.064$). A knowledge-tracing model (SAKT) on the seven TOEIC exam sections predicts next correctness only modestly better than a baseline that knows only how hard each section usually is (AUC lift $+0.051$; CI $[+0.045,+0.058]$), and that mastery signal is nearly independent of behavior styles (ARI$=0.007$). Behavioral clustering here describes study styles and engagement, not knowledge gains.