# Research Papers — 2026-07 · bundle page 30 of 222 > Full-text markdown of papers #1451–1500 of 11075 published in 2026-07, newest first. AI-readable mirrors of the full 11075-paper 2026-07 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-07/page/30 [← Previous bundle](https://aipod.dev/papers/archive/2026-07/page/29.md) [Next bundle →](https://aipod.dev/papers/archive/2026-07/page/31.md) ### Registration-Grounded Spectral Fusion for Unregistered WLI/NBI Endoscopic Lesion Segmentation Authors: Pengyu Jie, Wanquan Liu, Rui He, Pengcheng Li, Weiping Wen, Deyu Meng, Junwei Han, Chenqiang Gao Published: 2026-07-29 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.26395v1 White-light imaging (WLI) and narrow-band imaging (NBI) provide complementary views of endoscopic lesions, but their paired observations are often spatially misaligned due to viewpoint changes, tissue deformation, and sequential handheld acquisition. This makes direct WLI/NBI fusion prone to mixing non-corresponding regions and may even degrade segmentation around lesion boundaries. To address this problem, we propose a reliability-aware complex-domain fusion framework for paired-but-unregistered WLI/NBI lesion segmentation. The framework first establishes topology-regularized feature correspondence and further estimates where the cross-modal correspondence is reliable. Guided by this reliability, the model selectively fuses WLI and NBI features in a learnable complex representation. In this representation, WLI-derived cues mainly provide appearance-related magnitude responses, while NBI-derived cues provide structure-sensitive phase responses. Unlike conventional real-valued or symmetric multimodal fusion, the proposed method explicitly models the different roles of WLI and NBI and suppresses unreliable cross-modal interaction in locally mismatched regions. Experiments on paired WLI/NBI endoscopic datasets show that the proposed reliability-aware registration grounding and complex-domain fusion consistently improve lesion segmentation performance. Role-reversal and module ablation studies further validate the necessity of both the modality-role design and reliability-guided cross-modal interaction. --- ### CaM-Wolf: Causal-Aware Multimodal Agents for Social Deduction Games Authors: Zheng Zhang, Nanjie Yao, Jiarui He, Deheng Ye, Peilin Zhao, Hao Wang Published: 2026-07-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.26393v1 Social deduction games (SDGs) such as Werewolf have become challenging testbeds for AI agents. These games require complex social skills such as reasoning, deception, and collaboration. While recent advances in large language models (LLMs) have driven significant progress in SDG agents, current approaches are predominantly text-based, overlooking the multimodal nature that is fundamental to human social interaction. To bridge this gap, we introduce CaM-Wolf, the first SDG agent that integrates multimodal perception and generation. CaM-Wolf processes video inputs from other players, employs a causal-aware Reasoner trained via reinforcement learning to establish logical chains between observable behaviors and hidden roles, and presents itself through an animated avatar. Our experiments and user study show that CaM-Wolf achieves superior agent gameplay performance and enhances the quality of human-AI interaction. This work represents a significant advancement towards creating more human-like AI agents capable of participating in nuanced social dynamics. Our code is available at https://3dagentworld.github.io/avatar_wolf. --- ### Q-Steer: Action-Value Guidance for Molecular Policy Optimization Authors: Xinyu Wang, Jinbo Bi, Minghu Song Published: 2026-07-29 Categories: cs.LG, q-bio.BM arXiv: https://arxiv.org/abs/2607.26391v1 Oracle-limited molecular optimization gives reward only after a complete molecule is generated, while each rollout requires many local next-token decisions. This delayed-feedback interface makes molecular policy optimization myopic: an optimizer can learn that a molecule was good without knowing which intermediate actions made it good. We introduce Q-Steer, a rollout-time action-value steering primitive for molecular language models. Q-Steer uses an offline-trained and frozen prefix-action value scorer, PAVS-Q, that estimates the downstream reward of taking a candidate next token under a partial SMILES prefix, then adds a normalized value bonus to sampling logits. The optimizer update rule and online oracle budget are unchanged; the claim is fixed-online-oracle performance, not equal total compute. On PMO23 with a fixed 10,000-call online budget, complete factorial studies across two molecular language-model backbones and four optimizers show that Q-Steer improves mean valid-unique score in all eight backbone-optimizer cells, with positive macro mean-score gains between +0.033 and +0.049 and 18-20 task wins per cell. Mechanism controls show that action identity matters: prefix-broadcast values are nearly neutral, while shuffled action values harm performance. These results support Q-Steer as a reusable rollout-time action-value wrapper that improves average molecular optimization reward across optimizer families and policy backbones without changing the online oracle budget. --- ### Misalignment Has a Personality: A Big Five Account of Emergent Misalignment Authors: Hasibur Rahman, Smit Desai Published: 2026-07-29 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.26389v1 Fine-tuning a language model on data containing a narrow flaw, such as insecure code or incorrect mathematical answers, can cause broad misalignment through a mechanism that remains debated. We provide an interpretable account: in the models and corpora we study, misalignment behaves like a shift in personality. Prior work extracts activation directions for character traits from a single binary contrast, which can separate or steer behavior without establishing a calibrated scale. We instead extract personality vectors for the Big Five using a graded, three-level intervention and validate them on two open-weight models. The three levels are linearly ordered, with Cohen's d values of up to 6.2; the vectors transfer zero-shot and trait-specifically to an independent corpus; and their effects are strongest within a middle-layer band. Applied to training data, the vectors reveal that misaligned corpora across eight domains share a common Big Five signature: lower agreeableness and conscientiousness, together with higher extraversion and neuroticism. This signature is recovered by both models with a correlation of r = 0.94. Fine-tuning imprints the same profile, shifting the model's generations along the corresponding signature, with r = 0.83 using activation-based measurements and r = 0.90 using a text-based judge, while also shifting internal activations with r = 0.69. The same vectors characterize sycophancy as high extraversion and low conscientiousness rather than excess agreeableness, a distinction that a single direction cannot capture. Calibrated personality vectors transform an opaque safety phenomenon into a human-legible diagnostic profile. --- ### Collusion with Competitive Marginals: Price-Level Audits Are Blind by Construction Authors: Xin Xu, Chengrui Wu, Jiayu Lu, Kaizhen Tan, Siru Tao, Hanzhe Hong Published: 2026-07-29 Categories: cs.GT, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2607.26385v1 Empirical work on algorithmic collusion asks one question of the data: are prices supracompetitive? We show this can be answered "no" by a conspiracy that is nonetheless profitable. Consider bidding agents that couple only through the joint distribution of their unexplained bid components, leaving every agent's own bid law exactly at the competitive law. Any test whose input is a single agent's price or bid history then has power exactly equal to its false-positive rate, for every coupling strength up to comonotonicity. The published detection methodology is therefore blind to this conduct by construction rather than underpowered, and no sample size repairs it. Three empirical results follow. First, the mechanism appears in real language-model agents: twenty models from nineteen independent developers, three deployment prompts each, show residual correlation of $+0.053$ between two deployments of one model against $+0.0001$ across models, with a 95% interval clustered by developer of $(0.030, 0.078)$, under an auditor that sees every order feature and is fitted out of sample. Second, the coupling falls monotonically as sampling temperature rises ($p=0.002$), turning a deployment parameter into a candidate mitigation. Third, on 24 days of Ethereum block-building auction data covering 77,684 bids from 39 bidders, the honest population of bidder pairs is itself so dependent that a screen held at a 5% false-positive rate must sit above a floor of $+0.50$ to $+0.81$, which is 20 to 32 times the family-wise sampling threshold and does not fall as the audit window grows. Since lawful multi-identity operation and conspiracy are behaviourally indistinguishable here, the tractable regulatory target is not detection but counting: resolving 40 bidding identities into 23 operators raises the Herfindahl index by 247.5%, and adding behavioural clusters from public bid streams reaches 324.5%. --- ### Zero-Fi: Zero-Shot Wi-Fi-Based Human Activity Recognition via Contrastive Signal-Language Alignment Authors: Yitong Shen, Cheng Guo, Peiliang Wang, Jingzhe Zhang, Yi Sheng, Haopeng Zhang, Hongfei Xue, Yili Ren Published: 2026-07-29 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2607.26381v1 Wi-Fi-based human activity recognition has advanced substantially, but most existing methods assume a closed set of activities and require labeled Wi-Fi samples for every target class, limiting their ability to recognize unseen activities. We present Zero-Fi, a contrastive signal-language alignment framework for zero-shot Wi-Fi-based human activity recognition. Zero-Fi learns unified representations from complementary Wi-Fi signal features and aligns them with the semantic representations of natural-language activity descriptions in a shared embedding space. This cross-modal alignment enables Zero-Fi to recognize new activity classes without requiring labeled Wi-Fi samples or model adaptation for those classes. Experiments on large-scale public benchmark datasets demonstrate effective zero-shot recognition of held-out activity classes, highlighting the potential of signal-language alignment to extend Wi-Fi sensing beyond predefined activity classes. --- ### (Im)Paired Programming: Coding Agents Improve Productivity but Harm Understanding Authors: Nishant Balepur, Connor Baumler, Valerie Chen, Eunsol Choi, Rachel Rudinger, Jordan Lee Boyd-Graber Published: 2026-07-29 Categories: cs.CL, cs.HC arXiv: https://arxiv.org/abs/2607.26375v1 Coding agents (e.g., Cursor) improve developer productivity by optimizing task completion, but shifting users from writing code to prompting and reviewing may harm their understanding, impeding oversight, learning, and communication. To probe this, we have 54 students create a website with one of two AI systems: an agent that edits user code; or a chatbot where users write code alone or adapt generic code snippets. We test understanding via comprehension questions and a task where users extend their code without agents, showing: (1) While agents aid initial task completion, they harm users' code comprehension and thus do not prepare users to extend their code; (2) Low-effort agent interaction types, like copy+paste prompts and auto-accepted edits, are linked with lower comprehension; and (3) Despite self-reported weaker understanding, users still prefer coding agents because they are quick and easy to use. While users stay in the loop for coding workflows, understanding should not be forgotten. Towards this goal, we distill our analyses into future research directions for coding agent developers: dissuading low-effort prompting, creating readable code, and promoting active engagement. --- ### Self-Adaptive Learning and Model Predictive Control for Tracking Unknown Dynamics with No Regret Authors: Atharva Navsalkar, Hongyu Zhou, Vasileios Tzoumas Published: 2026-07-29 Categories: cs.RO, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2607.26370v1 We propose a self-adaptive online learning for control method for tracking unknown target dynamics. The target dynamics can exhibit switching behavior, particularly, a mixture of structured, random, and/or adversarial motion. Such challenging target tracking scenarios arise in applications of dynamic mapping, traffic control, and pursuit evasion, where robots need to track, pursue, or avoid collision with moving landmarks, objects, humans, etc., whose dynamics are unknown. Our method simultaneously learns multiple predictors from scratch, via self-supervised, one-shot, and computationally efficient learning, and adaptively selects the best one to match the observed target behavior. The method enjoys finite-time near-optimality guarantees in expectation, characterized as a function of the learning error of the target dynamics and the frequency that the target dynamics switch. In the absence of both error and switching, the method asymptotically matches the optimal non-causal control policy that knows a priori the target dynamics, i.e., the method enjoys no regret in expectation. In the presence of learning errors and switching, the method degrades gracefully, \eg when there are errors and no switching, the average regret is proportional to the average learning error and switching times. To prove these guarantees, a novel technical approach is required compared to the existing works that employ RFF-based online learning. We validate our method in Crazyflie simulations and hardware experiments, across target trajectories that vary from structured to random to adversarial, in comparison to non-stochastic, kernel-based, and neural-network-based methods for online learning. --- ### ClockRoPE: Random Fourier Rotations for Temporal Routine Modeling Authors: Yiwen Chen, Joshua Ainslie, Krzysztof Choromanski, Xiang Gao, Su-Lin Wu, Yiping Yuan, Qian Sun Published: 2026-07-29 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.26369v2 Rotary Position Embedding (RoPE) has been widely adopted in transformer-based large language models. However, its log-linear frequency schedule, originally designed to produce long-term attention decay, limits its adoption in domains with more complex distance-correlation patterns, such as temporal periodicity in sequential recommendation. We investigate the expressiveness of general query/key rotations and find that any normalized continuous positive-definite attention modulation function can be approximated by random rotations induced by its own Fourier transform, which we term Random Fourier Rotations. Building on this theory, we propose ClockRoPE for routine modeling in sequential recommendation, where rotation frequencies are derived from periodic attention modulation functions. In online A/B tests, ClockRoPE demonstrates consistent improvements in valued engagement metrics, and has been successfully deployed in production-scale generative retrieval system at a major video-sharing platform. --- ### ClockRoPE: Random Fourier Rotations for Temporal Routine Modeling Authors: Yiwen Chen, Joshua Ainslie, Krzysztof Choromanski, Xiang Gao, Su-Lin Wu, Yiping Yuan, Qian Sun Published: 2026-07-29 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.26369v1 Rotary Position Embedding (RoPE) has been widely adopted in transformer-based large language models. However, its log-linear frequency schedule, originally designed to produce long-term attention decay, limits its adoption in domains with more complex distance-correlation patterns, such as temporal periodicity in sequential recommendation. We investigate the expressiveness of general query/key rotations and find that any normalized continuous positive-definite attention modulation function can be approximated by random rotations induced by its own Fourier transform, which we term Random Fourier Rotations. Building on this theory, we propose ClockRoPE for routine modeling in sequential recommendation, where rotation frequencies are derived from periodic attention modulation functions. In online A/B tests, ClockRoPE demonstrates consistent improvements in valued engagement metrics, and has been successfully deployed in production-scale generative retrieval system at a major video-sharing platform. --- ### Diagnosing Fine-Grained Inconsistency Classification in Financial Disclosure Text Authors: Aman Kumar, Lasitha Vidyaratne, Dipanjan D Ghosh, Arnab Chakrabarti, Ahmed K Farahat Published: 2026-07-29 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.26368v2 Financial disclosures may contain numerical, temporal, referential, factual, and policy inconsistencies that require different evidence and reasoning to diagnose. We study \emph{fine-grained inconsistency classification}: given a passage known to contain a conflict, the goal is to identify its type among 11 categories. Using a fixed snapshot of the synthetic SBID-FD benchmark, we compare frozen and fine-tuned encoders, evidence-augmented classifiers, prompted large language models, and LoRA-adapted generative models under a shared evaluation protocol. Task-specific adaptation yields large improvements over frozen representations, and a fine-tuned 300M encoder performs competitively with substantially larger prompted and adapted models. We further study whether localizing the conflicting claims improves classification through matched predicted-span, reference-span, and distractor-span conditions. The results show that automatically extracted evidence provides additional signal but recovers only part of the benefit obtained from reference spans. Per-class and confusion analyses further reveal that some inconsistency types are especially sensitive to localization quality, whereas others remain difficult even when the relevant evidence is supplied. These findings identify evidence localization and fine-grained type discrimination as distinct challenges and show that compact supervised encoders are strong baselines for this task. --- ### Diagnosing Fine-Grained Inconsistency Classification in Financial Disclosure Text Authors: Aman Kumar, Lasitha Vidyaratne, Dipanjan D Ghosh, Arnab Chakrabarti, Ahmed K Farahat Published: 2026-07-29 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.26368v1 Financial disclosures contain numerical claims, temporal statements, entity references, policy commitments, and risk descriptions that may conflict in qualitatively different ways. Detecting a conflict is only the first step: review workflows may also need to determine its type, since numerical, temporal, referential, factual, and normative inconsistencies require different evidence and downstream checks. We study this problem as fine-grained inconsistency classification. Using a fixed 5,940-instance snapshot of SBID-FD, a synthetic financial-disclosure benchmark with 11 inconsistency labels and paired reference evidence spans, we compare frozen embedding classifiers, fine-tuned encoders, evidence-augmented classifiers, prompted large language models, and LoRA-adapted generative models under a shared evaluation protocol. A fine-tuned 300M encoder reaches 61.9% accuracy, compared with 61.5% for a LoRA-adapted Qwen3.5-9B model and 61.3% for GPT-5.4. Because these systems differ in architecture, supervision, training objective, and input format, we interpret this as a practical efficiency result for compact supervised encoders rather than a controlled conclusion about model scale. Supplying gold evidence spans improves the fine-tuned encoder to 65.3%, whereas automatically predicted spans recover a meaningful but incomplete share of that gain, indicating that localization quality remains a bottleneck. Class-level analyses show that Referential inconsistencies are especially sensitive to localization quality, while Factual and Logical inconsistencies remain difficult even when the relevant evidence is provided. Together, the oracle, distractor, and per-class analyses separate localization errors from residual type-discrimination errors, indicating that progress requires both stronger evidence extraction and better reasoning over closely related inconsistency categories. --- ### Exploring Structures in Physics Problems: Can AI Agents Discover Statistical Mechanical Mappings? Authors: Wanyu Zhao, Wanbing Zhao Published: 2026-07-29 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.26367v1 An important skill in theoretical physics is to recognize when a new problem can be transformed into a known model. We study this skill as an AI-agent task: can LLM-based agents discover statistical mechanical mappings from a raw partition function to a tractable representation? To probe this question, we introduce StatMechBench-v0, a benchmark of six Ising-type problems covering transfer-matrix methods, gauge-removable disorder, and planar/Pfaffian structure. We evaluate a simple propose-verify-revise agent across multiple LLMs and problem phrasings. The results show that numerical feedback often helps agents repair code and recover correct partition functions. However, agents can also pass the numerical checks while misidentifying the underlying tractable class or understating computational complexity. This both reveals limitations in current LLM reasoning and calls for a verification stack that goes beyond numerical agreement, incorporating, for example, symbolic checks and structural invariants. Our study provides an early evaluation and design directions for AI agents aimed at structural discovery in theoretical physics. --- ### Post-Training at the Edge of Detectability: A Game-Theoretic Approach to Fine-Tuning Authors: Keegan Harris, Brian W. Lee, Ian Waudby-Smith, Philip Amortila, Nika Haghtalab, Michael I. Jordan Published: 2026-07-29 Categories: cs.LG, cs.AI, cs.GT arXiv: https://arxiv.org/abs/2607.26358v2 Reinforcement learning (RL) fine-tuning is widely used in language model training to improve performance on a target task while limiting drift from a reference policy. A standard way to balance this trade-off is via a KL-regularized RL objective, although this formulation does not by itself provide a principled way to set the regularization coefficient. In practice, the coefficient is typically chosen heuristically or via hyperparameter search, which can lead to unnecessary overhead in training cost or undesirable reward-retention trade-offs. We instead propose a game-theoretic framework that gives this trade-off an explicit statistical interpretation. Specifically, we study a sequential game in which an agent chooses a policy to maximize cumulative reward while a monitor observes policy outputs over time and tests for deviations from the reference policy. Although not originating from the same perspective, we show that the resulting equilibrium policy can nonetheless be expressed as the solution to a KL-regularized RL problem for an optimal regularization parameter that can be viewed as maximizing reward per unit of statistical distinguishability. Drawing on classical results from concave-convex fractional programming, we provide a principled method for learning this equilibrium coefficient via reduction to the KL-regularized RL objective, thus allowing for flexible integration into standard fine-tuning pipelines. In experiments with Qwen3-8B and Llama-3.2-1B, we show that our methods result in competitive reward-retention trade-offs in a continual learning setting, and illustrate how our framework may be used to audit API providers serving open-source models. --- ### Post-Training at the Edge of Detectability: A Game-Theoretic Approach to Fine-Tuning Authors: Keegan Harris, Brian W. Lee, Ian Waudby-Smith, Philip Amortila, Nika Haghtalab, Michael I. Jordan Published: 2026-07-29 Categories: cs.LG, cs.AI, cs.GT arXiv: https://arxiv.org/abs/2607.26358v1 Reinforcement learning (RL) fine-tuning is widely used in language model training to improve model performance on a target task while limiting drift from a reference policy. A standard way to balance this trade-off is via a KL-regularized RL objective, although this formulation does not by itself provide a principled way to set the regularization coefficient. In practice, the coefficient is typically chosen heuristically or via hyperparameter search, which can lead to unnecessary overhead in training cost or undesirable reward-retention trade-offs. We instead propose a game-theoretic framework that gives this trade-off an explicit statistical interpretation. Specifically, we study a sequential game in which an agent chooses a policy to maximize cumulative reward while a monitor observes policy outputs over time and tests for deviations from the reference policy. Although not originating from the same perspective, we show that the resulting equilibrium policy can nonetheless be expressed as the solution to a KL-regularized RL problem for an optimal regularization parameter that can be viewed as maximizing reward per unit of statistical distinguishability. Drawing on classical results from concave-convex fractional programming, we provide a principled method for learning this equilibrium coefficient via reduction to the KL-regularized RL objective, thus allowing for flexible integration into standard fine-tuning pipelines. In experiments with Qwen3-8B and Llama-3.2-1B, we demonstrate that our methods result in competitive reward-retention trade-offs in a continual learning setting, and illustrate how our framework may be used to audit API providers serving open-source models. --- ### High-Order Markov Blanket Discovery via a k-Order Relaxation of the Faithfulness Assumption Authors: Loong Kuan Lee, Ragavi Krishnamoorthy, Nico Piatkowski Published: 2026-07-29 Categories: cs.LG, cs.AI, stat.ML arXiv: https://arxiv.org/abs/2607.26357v1 The problem of learning the graphical Markov blanket (MB) of a variable from data has applications in many areas such as structure learning for Bayesian networks and Markov random fields, causal discovery, and feature selection. However, a common assumption most methods make is that the conditional independencies in the distribution imply the same separation in the graphical structure -- also known as the faithfulness assumption. Unfortunately, this assumption can be violated by higher-order dependencies such as XOR and parity-type relations, and -- on finite samples -- by empirical violations that, in extreme cases, even induce spurious dependencies absent from the true distribution. Therefore, in this paper we propose a "k-order" relaxation of the faithfulness assumption that captures parity type relationships between k+2 variables. We then propose a proof of concept algorithm called k-order Markov blanket (kOMB) that uses this relaxation for MB discovery. Finally, we empirically show how kOMB can recover the MB of a variable under both true and empirical violations of faithfulness. Code available at: https://github.com/lklee9/k-order-Markov-blanket --- ### Symphony of Bias: Exploring Gender Associations with Musical Instruments in Multimodal LLMs Authors: Farhan Farsi, Shayan Bali, Mohammad Heydari Rad, Negar Heidary, Donya Rooein Published: 2026-07-29 Categories: cs.CL arXiv: https://arxiv.org/abs/2607.26355v1 Large language models (LLMs) are increasingly embedded in everyday life and widely used for information seeking, raising concerns about their potential to perpetuate social biases and reinforce stereotypes. In this study, we investigate gender bias in LLMs through the lens of their associations with musical instruments. Building on social-science research on the cultural gender-typing of instruments, we introduce Symphony-Bias, a parallel multimodal dataset spanning text, vision, and audio. We evaluate ten multimodal models with diverse architectures and scales across 22 musical instruments, analyzing how they associate each instrument with three gender categories: {male, female, non-binary}, across three modalities: {text, vision, audio}. Our results show that 92\% of instrument-level outcomes align with prior social-science findings, with the harp and drums showing particularly consistent gendered associations across all evaluated models and modalities. We further find that alignment with social stereotypes is weakest in audio, stronger in vision, and strongest in text, suggesting that modality-specific representations can differentially amplify gendered associations with musical instruments.\footnote{The Symphony-Bias dataset will be publicly released upon acceptance of the paper.} --- ### A Generative Grammar Underlying the Voynich Manuscript, the Pastiche Hypothesis: Evidence from Large Language Models Authors: Nicolas Turenne Published: 2026-07-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20835v1 Background: The Voynich Manuscript is a fifteenth-century codex written in an unknown script whose content remains undeciphered. Previous studies suggest that its statistical properties resemble those of natural languages, while its illustrations - primarily plants - recall medieval herbals. Methods: We present a multidisciplinary analysis combining probabilistic modeling, phonetic decomposition, rare-event detection, and multimodal image analysis, based on a newly transliterated corpus. Word- and letter-level distributions are modeled using position-dependent probabilistic grammars, while phonetic patterns are compared across Indo-European, Semitic, and Asian languages. Image-text alignment methods based on large language models are applied to identify potential botanical correspondences. Results: The results indicate that Voynich symbols behave as letters rather than syllabic units, while word-length distributions resemble syllabic structures. Phonetic analyses show closer alignment with consonant-heavy languages such as Hebrew or Arabic than with Indo-European languages. Probabilistic modeling reproduces Zipf-like distributions and reveals extremely low probabilities for repeated initial-letter sequences, indicating a structured imitation of natural language. Image analysis suggests strong correspondences between Voynich plant illustrations and those found in Pseudo-Apuleius herbals from the Mediterranean tradition, consistent with an imitation of medieval medicinal books. Perspectives: These findings support the hypothesis that the Voynich Manuscript follows a structured generative system combining linguistic regularities and herbal knowledge, and demonstrate the value of integrating probabilistic and AI-assisted approaches in the analysis of historical manuscripts. --- ### AI or Real: Detecting Partially Altered Videos Under Resource-Constrained Environments Authors: Tamoghna Chakraborty, Md Nurul Absur, Sourya Saha, Saptarshi Debroy Published: 2026-07-28 Categories: cs.CV, cs.DC arXiv: https://arxiv.org/abs/2609.20263v1 The proliferation of generative video models has shifted the practical detection threat from fully fabricated clips to partially manipulated footages. Although modern detectors achieve strong accuracy using foundation backbones of 400M+ parameters, their resource footprint precludes edge deployment. In this paper, we present a lightweight full-frame detector for partially manipulated AI-generated video, designed for deployment on edge hardware without face-detection preprocessing. The system distills a DINOv2-Base teacher into a frozen MobileNetV3-Small student through a pipeline that combines temperature-annealed soft-label transfer, attention-diversity regularization, frame-level supervision, and a residual feature adapter that conditions ImageNet features for artifact detection. We additionally target two failure modes specific to the partial-manipulation regime: false positives on legitimate scene cuts, addressed through within-video temporal hard negatives; and threshold-level miscalibration on the dominant pure-real class, addressed through calibration-aware sampling. Evaluation on a 55,393-sample spliced test set across fake-frame ratios from 6.2% to 31.2% demonstrates the student model closing 58% of the gap to the DINOv2-Base teacher (AUC 0.766) while running at 3.65 ms per 16-frame clip on RTX A4000 with a 150.4 MB checkpoint compatible with edge-device memory and latency budgets. --- ### Generative Verification: Rethinking the Uncertainty Signal for Active Learning of Object Detection Authors: Licheng Zhang, Zheng Gong Published: 2026-07-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20262v1 Nearly every acquisition function for active object detection shares one arrangement, in that the model being improved is also the model being interrogated. We depart from it. In generative verification an independent generative model re-derives the label of a detection from the pixels inside its predicted box, and the disagreement between the two becomes the acquisition signal. Two properties follow from the arrangement itself rather than from any tuning. A displaced box, a box on background and a correct box carrying the wrong label all yield a crop that fails verification, so the failure modes arrive already combined in one scalar and the hand-weighted classification and localization terms of existing criteria are no longer needed. And because the verifier never observes the detector confidence, confidently wrong detections score highest, although a self-derived signal reads them as uninteresting and they are the costliest to leave unlabeled. We build the verifier as a conditional diffusion model whose diffusion target is a label representation rather than an image. Its reverse process is stochastic, so repeated generations return a distribution whose concentration reports how firmly the evidence determines the label, where a classifier returns a single point estimate. On PASCAL VOC and MS-COCO the signal outperforms output-uncertainty, feature-geometry, perturbation and ensemble criteria, gaining about one mAP50 point per round on MS-COCO, with its largest margins in the early rounds where confident detector errors are most common. --- ### When AI Agents Commit: Cognitive Serializability Across Data, Evidence, Policy, and Authority Authors: Jun He, Deying Yu Published: 2026-07-28 Categories: cs.AI, cs.DC arXiv: https://arxiv.org/abs/2609.20261v1 Autonomous agents derive concrete mutations from database reads, retrieved evidence, policy, beliefs, and delegated authority. Those inputs may change while reasoning is in progress. Database isolation orders the submitted transaction; agentic transaction processing determines whether a proposal satisfies an executable contract. Neither guarantee establishes a common valid point for the mutation and its derivation inputs unless the contract represents the relevant predicates. Typed dependency tokens distinguish content integrity from applicability, and trusted mediation captures the values exposed to reasoning. Under strict Cognitive Serializability, committed effects admit a serial order and a logical event at which every value exposed to derivation is unchanged. The fences last until the runtime event that realizes the sealed durability domain. The weaker Effect-Compatible Cognitive Admission recertifies an effect against a simultaneously held current dependency vector and current policy without claiming to serialize the original stochastic derivation. TCT combines immutable versioned executable definitions, registry-derived authority plans, sealed envelopes, guard-first commit transactions, post-seal envelope- and witness-bound grants, co-committed receipts, idempotent grant finalization, and receipt-driven epistemic reconciliation. Complete registered footprints and a single growing phase induce an acyclic lock-point order over local guards and incompatible external reservations. The corresponding results give serializability conditions and an observational-equivalence boundary for zero-error soundness and positive progress. A falsification suite tests the implementation obligations: the prototype prevented all injected anomalies and added 3.22 ms mean commit overhead. --- ### Lens: Bringing the Right Semantic Perspective into Focus for Training-Free Multimodal Representation Learning Authors: Xinran Liu, Shouqian Shi, Yixian Chen, Ruizhi Chen, Xin-Wei Yao, Sheng Zhong Published: 2026-07-28 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.20252v1 High-quality representations are essential for a wide range of downstream tasks. Dedicated embedding models are explicitly optimized for representation learning, yet their training data are often more limited in scale and diversity than the massive corpora used to pretrain modern large language models and multimodal large language models. Large-scale pretraining and instruction following enable autoregressive models to select relevant evidence, integrate multimodal information, and infer semantics under different task perspectives, creating a distinctive opportunity for training-free representation learning. However, our analysis reveals that existing semantic-elicitation methods do not reliably orient the extracted states toward the semantic perspective required by the downstream task. Consequently, the resulting representations often remain dominated by salient input content. We characterize this problem as semantic perspective misalignment and propose Lens, a training-free framework that makes representation readout task-directed. Semantic Perspective Anchoring associates the task-required perspective with a task-specific readout phrase, specifying the interpretive role of the positions later used for extraction. Contextualized Phrase Readout places the same phrase after the complete input and aggregates its token states, combining full-context access with the anchored perspective. The resulting representation reflects task-conditioned evidence integration and inference rather than a generic summary of salient content. Without parameter updates, architectural modification, or reranking, Lens achieves an overall Precision@1 of 63.9 across all 36 MMEB datasets, outperforming the closest same-backbone training-free embedding baseline by 10.2 points. --- ### Accuracy Is Not Enough: A Cross-Architecture Audit of Demographic Bias in Deep Knowledge Tracing Authors: Dang Quang Minh, Nguyen Dung Son, Nguyen Huu Loi, Truong Viet Vu, Nguyen Thai Anh Published: 2026-07-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.20249v1 Deep knowledge tracing (DKT) models implicitly decide which students an adaptive system believes have mastered a skill, yet almost all evidence on their demographic fairness comes from Bayesian knowledge tracing; the deep models that power modern systems have received no comparable cross-architecture audit. We close this gap: four architectures (DKT, DKVMN, SAKT, AKT) trained under three regimes (standard, reweighting, adversarial) on two public datasets with demographic metadata, Eedi (15.9M interactions) and OULAD (167k after preprocessing), evaluated with ABROCA, student-level bootstrap confidence intervals, and permutation tests addressing recent critiques of fairness-metric instability. Three findings emerge. (i) Bias is real but context-dependent: every architecture shows a significant socioeconomic ABROCA on Eedi (0.018-0.023, $p<0.005$), with per-group AUC lower for economically disadvantaged students, while gender bias is significant on OULAD for three of four architectures after multiplicity correction yet negligible on Eedi. (ii) The most accurate architecture is the most biased: AKT gains about 4 AUC points from item-level Rasch embeddings and shows the largest socioeconomic ABROCA, exceeding every other architecture under a paired bootstrap ($p\leq0.002$); ablating only the Rasch embeddings removes the accuracy gain and the excess bias together. (iii) Standard mitigation is unreliable: reweighting and adversarial debiasing leave ABROCA essentially unchanged in every configuration that preserves accuracy, even though the adversary is pinned at chance at full reversal strength and a weak-strength positive control rules out a dead probe. --- ### Distance to Class Prototypes: Active Learning for Object Detection Authors: Licheng Zhang, Zheng Gong Published: 2026-07-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20248v1 Deploying a deep object detector in a new setting is limited less by architecture than by the cost of annotating data from that setting. Active learning lowers the cost by choosing which images to label, and the choice is only as good as the signal used to score an unlabeled image. That signal is usually the class posterior, which is cheap but poorly calibrated, or the disagreement across several models or several stochastic passes, which is better but multiplies inference over a pool far larger than the labeled set. We propose a signal richer than the posterior yet still read from one forward pass of one network. A supervised contrastive term added to the training objective shapes a per-object embedding space in which distance encodes class membership, and an unlabeled detection is scored by how far it lies from the region occupied by its predicted category, weighted by its confidence. The criterion needs no ensemble, no auxiliary predictor and no repeated inference, and its entire cost is 2.89M parameters, an increase of 8.3% over a bare detector. On PASCAL VOC and MS-COCO it beats the posterior of the same detector in every round in which a selection is made, by up to 1.08% mAP50 against run to run deviations of 0.02% to 0.18%, and it stays competitive with ensemble and Monte Carlo dropout criteria costing three to fifty forward passes per unlabeled image. Experiments use the single-stage detector under which the compared criteria report their results, so that the selection decision is isolated from the strength of the detector. --- ### SAGE-Yoga: Multi-Cue Learning for Yoga Pose Classification and Joint-Level Correction Authors: Hung Le Chi, Khanh Minh Huynh, Long Nghia Tran Pham, Tan Phuc Huynh, Trong-Thuan Nguyen, Minh-Triet Tran Published: 2026-07-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.20245v1 Automated yoga analysis requires both accurate pose classification and interpretable feedback on pose execution. However, existing methods often rely on a single visual prediction, struggle to distinguish visually similar poses, and treat pose classification and correction as separate tasks. To address these limitations, we propose SAGE-Yoga, a unified coarse-to-fine framework for yoga pose classification and joint-level correction from a single RGB image. Inspired by how yoga instructors assess posture using multiple complementary cues, SAGE-Yoga first employs a bagging-based ensemble of complementary visual backbones to generate a ranked set of candidate pose classes. Additionally, a margin-based gating mechanism preserves confident visual predictions while invoking geometric verification only for ambiguous cases. Moreover, once the final pose class is determined, SAGE-Yoga retrieves a medoid reference pose and compares the observed joint angles with class-specific distributions to identify misaligned joints. Finally, these deviations are translated into actionable corrective feedback. Empirically, experiments on the Yoga-82 dataset show that the visual ensemble achieves 89.0% Top-1 accuracy, while the complete framework improves performance to 90.7% Top-1 accuracy and 90.1% Macro-F1. These results demonstrate that combining complementary visual evidence with selective geometric verification improves fine-grained pose classification while enabling interpretable, joint-level correction. --- ### Scene-Q: Confidence-Aware Coarse-to-Fine Querying of 3D Scenes with Selective VLM Reasoning Authors: Juno Kim, Yesol Park, Hye-Jung Yoon, Byoung-Tak Zhang Published: 2026-07-28 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2609.20235v1 Indoor mobile robots require open-vocabulary scene understanding that grounds natural-language queries in a consistent 3D map. Many existing systems ultimately rely on cosine-similarity retrieval with contrastive image--text encoders, which is efficient but brittle when labels are near-synonymous or multiple similar instances appear. We present Scene-Q, a confidence-aware coarse-to-fine querying framework that normalizes encoder scores with temperature scaling and selectively invokes a reasoning VLM only for low-confidence cases. High-confidence queries are answered by fast retrieval, while ambiguous ones are reranked over a small top-K candidate set using the original multi-view images and instance bounding boxes, enabling context-aware disambiguation at low cost. Scene-Q improves open-vocabulary 3D instance segmentation on ScanNet200 and natural-language 3D instance retrieval on real-world reconstructions, with the largest gains on spatial and relational queries while keeping a substantial fraction of queries on the fast path. --- ### Before the Warning Comes Too Late: Incremental Phone-Scam Detection from Speech Authors: Khang Nhat Hoang Vo, Anh Trac Duc Dinh, Tai Tien Ta, Tho Quan Published: 2026-07-28 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.20223v1 We study weakly supervised incremental telecom fraud detection from raw telephone audio, where training provides only conversation-level labels and predictions must be updated before a call ends. We introduce StreamFraudNet, which processes incoming audio through overlapping bounded-context windows using a frozen self-supervised speech encoder, recurrent temporal modeling, and learned aggregation of latent window scores. On a controlled English benchmark, StreamFraudNet achieves a ROC--AUC of \(0.9953\), significantly outperforming acoustic and mean-pooling baselines while remaining competitive with strong global temporal models. The model produces its first prediction after 10 seconds of audio, updates every 2 seconds, and operates faster than real time on the evaluated server hardware. Ablations identify recurrent temporal context as the principal contributor to performance. These results demonstrate that fraud risk can be scored incrementally from raw speech without transcripts or temporal annotations, while highlighting the need for latency-aware training to improve early prediction. --- ### Is It Still Worth Training a Classical Model in the Era of LLMs? A Crossover Benchmark on Tabular Data Authors: Kaihua Ding Published: 2026-07-28 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.20218v1 Large language models can label a tabular row from a plain-English description with no training - a capability now shipping in mainstream spreadsheet tools such as Microsoft Copilot in Excel and Anthropic's Claude for Excel - raising a practical question for the many business prediction problems where labels are expensive: should you prompt a frozen LLM, or collect data and train a model - and if so, how much data? We quantify the answer with the labeled-data crossover N*, the training-set size at which a trained classical model's learning curve overtakes a frozen LLM's training-free (and therefore flat) error. Aggregating 126 independent student evaluations of small GPT models under eight prompting configurations across 18 tabular datasets, paired with authoritative power-law learning curves for six classical model families, we find that training wins fast: even given an oracle choice of its best prompt configuration, a trained classical model beats the small frozen LLM using no more labeled data than is already on hand in 86% of cases, and wins by the smallest labeled subset we evaluate in 40%, with the observed crossover at a median of ~6% of the training set. In-context few-shot examples do not behave like training - error versus shot count does not follow a power law - and the same protocol re-run by independent implementers varies with a coefficient of variation of 0.148. A controlled probe indicates the LLM depends on recognizable feature-name semantics, which plausibly makes our crossover a conservative estimate (we do not claim memorization). For a typical business table, the evidence is clear: collect a few hundred labels and train a gradient-boosted model. --- ### Transformer fault diagnosis using an efficient simulation-driven variational quantum classifier with domain-aware feature encoding Authors: Huy Hoang Le, Ba Tu Phung, Dai Huynh, Kim-Anh Nguyen Published: 2026-07-28 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2609.20214v1 Early transformer fault diagnosis is challenged by nonlinear dissolved-gas interactions, overlapping fault signatures, and limited labeled data, while practical deployment further requires reliable performance under realistic computational constraints. This paper presents a simulation-driven modeling framework for dissolved gas analysis-based transformer fault diagnosis, in which a carefully engineered variational quantum classifier (VQC) is employed as the computational core and systematically analyzed through simulation. The framework integrates domain-aware feature modeling derived from Duval geometry with a lightweight two-qubit quantum representation, enabling nonlinear gas-interaction effects to be captured within a shallow parameterized circuit. A hybrid ZX-YY quantum feature map is designed to model non-commuting feature interactions, while a full-entanglement EfficientSU2 ansatz provides adequate expressive capacity under strict resource limits. Model behavior is evaluated using a comprehensive simulation pipeline including noise-aware circuit emulation, cross-dataset validation, and limited hardware-in-the-loop execution, allowing key effects of circuit depth, noise, and optimization strategy to be examined. Simulation results on benchmark dissolved-gas-analysis datasets demonstrate high diagnostic accuracy, strong generalization capability, and robustness to realistic noise levels with minimal quantum resources. The results highlight the effectiveness of simulation-informed modeling for practical transformer diagnostic applications, offering a reproducible and resource-efficient pathway for evaluating quantum-enhanced fault diagnosis methods. --- ### YNU-HPCC at SemEval-2025 Task 11: Bridging the Gap in Text-Based Emotion Using Multiple Prediction Headers Authors: Hao Yang, Jin Wang, Xuejie Zhang Published: 2026-07-28 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.19238v1 This paper describes the participation of the YNU-HPCC team in subtask A of task 11, Bridging the Gap in Text-Based Emotion at SemEval-2025. Our best-performing system employs the RoBERTa (Robustly Optimized BERT Approach) model, an improved version of BERT that utilizes the Transformer encoder architecture. We enhanced the output head to allow the model to process one emotion simultaneously. We obtained the official ranking score (0.44), including results from all languages. The entire dataset was translated into English using Google Translate to facilitate subsequent processing. Through probabilistic and attention analyses, we found that (I) a single prediction head performs better than six heads predicting six emotions simultaneously, and (II) training on a uniformly translated English dataset yields better results than using the original dataset. The code is available at: https://github.com/BGWH123/Semeval-2025-task11. --- ### KDTwin: Task-Aware Knowledge Distillation for Lightweight Multi-Task Driving Scene Segmentation Authors: Huy Che, Minh-Khoi Do, Dinh-Duy Phan, Duc-Khai Lam Published: 2026-07-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.18955v1 Efficient perception models are essential for real-time autonomous driving, where accuracy and computational cost must be carefully balanced. However, applying knowledge distillation to multi-task driving scene segmentation is challenging because drivable-area and lane segmentation exhibit different spatial characteristics and class imbalance. We propose KDTwin, a task-aware distillation framework for lightweight multi-task segmentation networks. The proposed method performs distillation at both the shared encoder and task-specific decoders. Encoder-level pairwise distillation transfers spatial relational knowledge to enhance the student's shared representation. For the decoders, we use a weighted loss for drivable-area segmentation and a boundary-aware loss for lane segmentation, enabling task-adaptive knowledge transfer without increasing inference complexity. Experiments on BDD100K show consistent improvements across the evaluated CNN-based and Transformer-based student models without increasing inference-time parameters or FLOPs. The results show that designing distillation objectives according to task-specific characteristics can effectively enhance multi-task segmentation performance for autonomous driving. The source code is available at https://github.com/chequanghuy/KDTwin. --- ### Optimized Fuzzy Logic Approach with the IEEE Key Gas Method for Diagnosing Power Transformer Faults Using Dissolved Gas Analysis Authors: Kim-Anh Nguyen, Huy Hoang Le, Ba Tu Phung Published: 2026-07-28 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.18133v1 Reliable transformer fault diagnosis is essential for maintaining power system stability. The IEEE Key Gas Method (KGM), a widely utilized approach in Dissolved Gas Analysis (DGA), exhibits limitations in addressing ambiguous data and ensuring high diagnostic accuracy. This study presents An enhanced model combining Fuzzy Logic with the IEEE Key Gas Method (FL-KGM) that introduces refined membership functions, optimized fuzzy rule sets, and a novel separation of CO and CO2 to eliminate diagnostic inconsistencies. By leveraging multidimensional gas ratio analysis and an adaptive classification framework, FL-KGM delivers superior fault identification and classification. Experimental validation utilizing real-world datasets demonstrates that FL-KGM achieves up to 98.6% accuracy, significantly outperforming KGM and other FL-based approaches. These findings elucidate the potential of FL-KGM in advancing transformer monitoring, enabling intelligent fault detection, and enhancing predictive maintenance strategies in modern power systems. --- ### Alignment Is All You Need: Instruction-Free Training for General Audio-Language Models Authors: Xuanru Zhou, Yiwen Shao, Jiahong Li, Dong Yu Published: 2026-07-28 Categories: cs.CL, cs.SD, eess.AS arXiv: https://arxiv.org/abs/2608.18132v1 Multimodal large language models (MLLMs) are typically built through a multi-stage pipeline consisting of cross-modal alignment, supervised fine-tuning (SFT), and preference optimization. This pipeline assumes that adapting an LLM to a new modality requires extensive task-specific supervision. However, pretrained LLMs already possess strong reasoning and instruction-following abilities. As LLMs evolve rapidly, an important question remains: can we efficiently transfer these capabilities to a new modality with minimal intervention, and is alignment alone sufficient for building a multimodal model? We introduce an Instruction-Free Alignment-Only large audio-language model (LALM) that keeps both the audio encoder and the LLM fully frozen, learning only a lightweight projector. Borrowing insights from AzeroS (1), we train on (audio, response) pairs from Self-Generated Data Construction, where an LLM expands captions into free-form responses without explicit task instructions. Across MMAU, MMAR, MMSU, and MMAU-Pro, our approach matches or surpasses heavily post-trained baselines using substantially less data. By keeping the LLM frozen, our model preserves its native instruction-following competence and can port seamlessly across model generations. Our results suggest that competitive MLLM can emerge from alignment alone, reducing multimodal extension to a lightweight projector-training problem that generalizes across modalities and adapts rapidly to each new LLM release. --- ### Task- and Session-Level Model Routing: A Common-Interface Hybrid Evaluation of Four Open-Source Routers Across Four Benchmarks Authors: Kiran N. Kumar, Santhosh K. Saminathan Published: 2026-07-28 Categories: cs.AI, stat.ML arXiv: https://arxiv.org/abs/2608.14641v1 Agentic systems increasingly delegate model selection to a router, yet open-source routers are usually evaluated with different tasks, candidate pools, and execution protocols, limiting direct comparison. We present a common measurement protocol and hybrid evaluation of four router implementations across RouterBench, BFCL v4, tau2-bench, and WebArena. We evaluate 290 frozen tasks against a locked matrix of 2,610 candidate outcomes. Three routers emit constant or near-constant tier assignments; only vLLM Semantic Router varies materially with prompt content, and it has the highest observed success rate on none of the four benchmarks. Always-Mid matches Aurelio exactly on three benchmarks and within 0.003 on the fourth. For vLLM, task-level superiority tests detect no task-specific advantage over a share-matched content-blind allocation; equivalence is established only on WebArena at the protocol-declared five-percentage-point margin. The results show that, under these configurations and controls, observed gains track selected-tier composition more closely than demonstrated task-specific targeting. Fixed-tier baselines and selected-tier distributions are therefore necessary controls in router evaluation; the findings are scoped to these configurations, candidate pool, and frozen benchmark samples, not to routing paradigms in general. --- ### BDIP-Net: Dual-Interaction Graph Learning for Property Prediction of Bilayer Materials Authors: An Vuong, Chen Zhao, Jin Hu, Shui-Qing Yu, Xintao Wu Published: 2026-07-28 Categories: cs.LG, cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2608.14640v1 Stacked bilayer materials exhibit rich stacking-dependent properties driven by the interplay between strong intra-layer bonding and weak inter-layer van der Waals interactions. The computational discovery of such materials is challenging because accurate structure generation typically relies on expensive DFT-based optimization, while existing machine-learning models often fail to explicitly distinguish different interaction types during property prediction. To address these challenges, we propose a machine-learning framework for efficient construction and property prediction of stacked bilayer materials. The framework employs a MatterSim-D3-based structural optimization workflow to generate DFT-quality bilayer structures from monolayer building blocks and stacking configurations at substantially reduced computational cost. For property prediction, we introduce BDIP-Net (Bilayer Dual-Interaction Potential Network), a graph neural network that explicitly models intra-layer and inter-layer interactions through interaction-specific potential representations and adaptive message fusion. We evaluate the proposed framework on BiDB, HetDB, and SAMBA, encompassing homobilayers, heterobilayers, and twisted bilayer systems. Results show that the MatterSim-D3-based workflow closely reproduces DFT-PBE-D3 optimized structures, while BDIP-Net consistently outperforms existing graph neural network and potential-based approaches for bilayer property prediction. --- ### Valid Per-Field Selective Risk Control for Document Extraction: Three Failure Modes, a Validity Ladder, and When Conditioning Pays Authors: Bhaskar Gurram Published: 2026-07-28 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.14639v1 Per-field accept/review with selective risk at most alpha -- accept a field only if the error rate among accepted fields is controlled -- is the trust contract document-extraction systems need, and the natural procedure silently violates it on real documents. On 13,859 genuine claude-sonnet-5 fields from 800 CORD receipts (49.0% correct) we diagnose three failure modes: document clustering (design effect 1.84-2.45), score-refit leakage (coverage 0.416 at risk 0.127, violating alpha=0.10 in 95% of splits), and a tie-mass pathology (a degenerate score collapses the threshold grid, 0.030 to 0.001). We organize the fixes as a validity ladder, guarantee form stated per tier. A fit/val split protocol restores expected-selective-risk control for a learned fusion: coverage 0.318 at risk 0.096 at nominal alpha=0.10, no tolerance band (production variant 0.326) -- an on-average point whose realized risk exceeds alpha in 47.5% of resplits, not a certificate. Mondrian Learn-then-Test with exact binomial tails yields per-group PAC certificates: field-iid 0.171 at risk 0.068, cluster-corrected 0.140, doc-iid 0.060 -- the only tier matching documents, honestly near-vacuous today. Support-bin, the pre-specified provenance taxonomy, wins every rigor tier on the sonnet CORD capture (p<1e-4, Bonferroni-corrected) -- a win that does not replicate on the same documents under haiku or qwen -- while on higher-accuracy corpora pooled thresholds win: conditioning helps exactly where pooled cannot certify, subsumed by a learned score elsewhere. A frozen-configuration confirmation on selection-untouched claude-haiku-4-5 held at both risk levels, and a blind three-annotator human-gold audit verifies the practical tier's accepted-set risk at 1.3% against its 10% budget (Fleiss' kappa=0.83; labels err one-sidedly pessimistic). Released Apache-2.0 with seed-pinned, regression-gated procedures. --- ### Randomly initialized autoencoders: fixed points and edge-of-chaos Authors: Leonid Berlyand, Roman Sarapin, Yitzchak Shmalo, Victor Slavin, Sasha Sodin Published: 2026-07-28 Categories: cs.LG, math.PR, math.ST arXiv: https://arxiv.org/abs/2608.14638v1 In this paper we study autoencoders, a special class of deep neural nets (DNNs) whose performance can be characterized via their fixed points. This perspective naturally raises questions of existence, stability, and basins of attraction of these fixed points. These questions are addressed via the contractive properties of autoencoders, and are closely related to the notion of edge-of-chaos. Edge-of-chaos (EoC) is an important notion in the theory of DNNs. It describes the critical regime separating ordered and chaotic signal propagation through a randomly initialized network. Initialization at or near this critical regime offers several theoretical and practical advantages, including stability of the network w.r.t. perturbations of the input. EoC was previously introduced for broad classes of neural networks using mean-field averaging methods. In this paper we modify the notion of EoC for the study of autoencoders. Specifically, we introduce local and global EoC for autoencoders that control local (small) and global (arbitrary) perturbations of the input respectively. The study of stability of autoencoders falls within the scope of nonlinear problems in Random Matrix Theory (RMT). Our analysis of local EoC is based on spectral techniques of RMT, whereas global EoC is studied by employing Sudakov-Fernique inequality for Gaussian processes. --- ### Early Cycle Charge Trajectory Generative Prediction and Full Life Cycle Health Management of Iron-Chromium Flow Batteries Based on FlowBD-E1 Authors: Suyang Zhuang, Zekun Jiang, Tianhang Zhou Published: 2026-07-28 Categories: cs.LG, stat.ME arXiv: https://arxiv.org/abs/2608.14637v1 Long-duration stationary energy storage requires batteries whose degradation can be detected before substantial capacity loss has accumulated. Iron-chromium redox flow batteries are attractive for this role because they use abundant and low-cost active species, yet their operation is shaped by slow chromium kinetics, hydrogen evolution, membrane crossover and electrolyte imbalance. These coupled processes gradually reshape the full charge voltage/current (V/I) trajectory, but most battery prognostic studies either focus on lithium-ion cells or compress ageing into scalar capacity and state-of-health (SOH) labels. Here we study an industrial 33 kW Fe-Cr redox flow battery and introduce FlowBD-E1, an early-cycle generative forecasting framework that predicts complete future charge V/I trajectories from only the first few cycles. The model combines a multi-scale convolutional encoder, a lifecycle Transformer and an age-aware FiLM decoder, and we compare three deployment strategies: single-step latent extrapolation (SLE), recursive latent forecasting (RLF) and teacher-forced updating (TFU). Using the first 9 of 289 cycles, RLF achieved a joint V/I mean absolute percentage error (MAPE) of 0.731% over the remaining lifecycle and produced SOH estimates below 1% MAPE. Ablation and independent-sequence tests showed that the age-aware generative architecture outperformed LSTM and TCN baselines and retained sub-percent errors under industrial validation. These results suggest that early-cycle trajectory generation can turn a short commissioning record into a long-horizon diagnostic signal for flow-battery management. --- ### Fractional Optimizers Meet Fractal Activation Functions: An Empirical Study of Multi-Scale Optimization in Neural Network Authors: Sebastian Raubitzek, Georg Goldenits, Sebastian Schrittwieser, Philip König, Kevin Mallinger Published: 2026-07-28 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.14636v1 Fractional optimization methods and fractal activation functions are two independent directions for improving neural network training. Fractional optimizers extend first-order optimization through fractional derivatives and memory effects, whereas fractal activations introduce multi-scale nonlinear representations based on self-similar Weierstrass- and Blancmange-type functions. Here, we investigate their interaction within a unified experimental framework. We evaluate fractional optimizer families on Ackley and Himmelblau benchmark surfaces, in standard form and with additive Weierstrass-type perturbations, and then in feed-forward neural networks with conventional and fractal activations on ten classification datasets. The comparison includes standard methods, regularization-style optimizers, explicit and adaptive memory-based fractional optimizers, and other representative literature methods. Overall, fractional optimization and fractal activations show useful but selective pairings. Regularization-style fractional scaling performs well with selected fractal activations in network training, while Grünwald--Letnikov memory is most relevant on perturbed surfaces. Adaptive memory improves plain memory substitution in several cases, supporting controlled fractional memory as a promising direction rather than a universal replacement. --- ### Belayer: Efficient Fault Tolerance for LLM Agentic RL Training Authors: Jiecheng Zhou, Qinghao Hu, Peng Sun, Xingcheng Zhang, Weiming Zhang Published: 2026-07-28 Categories: cs.DC, cs.LG arXiv: https://arxiv.org/abs/2608.14635v2 Large language model (LLM) agents are increasingly trained with reinforcement learning in long-horizon, sandboxed environments. Unlike conventional RL, agentic RL couples GPU-intensive rollout engines with stateful environment containers whose actions may produce visible side effects, such as file edits, command execution, and dependency installation. A single trajectory can span many rounds of gen- eration and environment interaction, so a component failure can discard completed work or expose the model to an environment state that is inconsistent with its context. However, existing systems lack efficient and correct recovery mechanisms for this distributed execution model. This paper presents Belayer, an efficient fault-tolerant system for LLM agentic RL training. Belayer handles failures in both rollout engines and environment execution while targeting low failure-free overhead. For scoped worker-local rollout failures, Belayer equips each pre-initialized shadow worker with a selective GPU-state reuse protocol that retains independently owned weights and raw KV-arena allocations after owner and GPU health checks, reinitializes worker-local state, and rebuilds request-specific KV contents from logged token prefixes. For environment failures, Belayer introduces full checkpoint and full restore to jointly capture and restore container file-system and runtime state, and coordinates the recovered environment with the LLM context to preserve prefix consistency. An adaptive policy opportunistically overlaps full-state checkpointing with natural LLM inference bubbles when the predicted interval is long enough. Empirical results show low measured overhead during failure-free training, a worker-recovery-time reduction of up to 42 times faster compared with a full engine cold start, and 1.5 to 3.5 times faster recovery from environment failures. --- ### Forecasting Side Effects of Activation Steering Authors: Chong Yong Ong, Alson Wei Jie Sim, Peixin Zhang, Jun Sun Published: 2026-07-28 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.11227v1 Activation steering modifies a language model by adding a learned direction to its hidden activations, enabling targeted behavioral changes without retraining. While effective, steering often produces unintended side effects on other behaviors, making it difficult to deploy safely. We therefore ask: can these side effects be forecasted before steering is applied? We answer this question by constructing a cross-effect matrix over a taxonomy of 67 behaviors across three open-weight language models. We find that side effects are common, structured, and often asymmetric, revealing interactions that cannot be explained by existing similarity-based heuristics. Despite this complexity, we show that side effects are largely predictable before steering is performed. Their magnitude depends primarily on the target behavior, while their direction can be forecasted from the model's unsteered representations with substantially higher accuracy than simple baselines. Our results demonstrate that activation steering has systematic and forecastable side effects, enabling proactive safety auditing and more informed deployment of steering interventions. --- ### SPOTting the Future: Lookahead Explanations for Deep Reinforcement Learning Authors: Tamar Gozlan, Claudia V. Goldman Published: 2026-07-28 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.09967v1 Deep reinforcement learning (DRL) agents achieve strong performance in complex environments, yet their decision-making processes remain difficult to interpret. We introduce SPOT (Sampling Policy Observation Tree), a novel model-agnostic, sampling-based framework for interpreting DRL policies. Given access to the policy and an environment simulator, SPOT constructs an interpretable finite-horizon tree by sampling actions and recursively simulating the resulting successor states. The tree provides an empirical representation of the policy's action preferences and their possible downstream evolution. We provide formal guarantees establishing SPOT's asymptotic recovery of the policy's unique most probable action and characterizing its disagreement behavior under high-entropy policies. We demonstrate SPOT in the SUMO-RL traffic-signal control domain. The case study illustrates how its tree-based representation can be used to inspect policy preferences, compare alternative future trajectories, and reveal downstream behaviors that are not visible through single-timestep feature-attribution methods. --- ### Projected climate memory and inherited warm-tail risk in accelerated European summer warming Authors: Mauricio Herrera-Marín, Alex Godoy-Faúndez, Diego Rivera Published: 2026-07-28 Categories: physics.ao-ph, cs.LG, stat.AP arXiv: https://arxiv.org/abs/2608.09966v1 European summer warming reflects interactions among background change, persistent ocean--land--circulation states, and same-season variability. We develop an empirical reduced-dynamics framework that decomposes regional summer indicators into inherited slow-state memory, its predictable component, and contemporaneous innovation. Projection-operator theory motivates the decomposition, implemented with finite causal filters, ridge-regularised prediction, and logistic risk models. Using ERA5-derived summer indicators for 28 IPCC AR6 European sub-regions over 1950--2024, with validation on 2006--2024, we find that Mediterranean-state memory improves mean summer-temperature prediction relative to trend-only and ARX baselines. The gain over ARX is modest, while moving-average, exponentially weighted, and tempered filters contain similar annual information, indicating that the data identify useful slow-state memory more robustly than a unique kernel shape. Predictable-state reconstruction is ridge-sensitive and therefore treated diagnostically rather than as a forecasting model. The strongest result concerns warm-tail risk. In parsimonious logistic models, high accumulated Mediterranean memory raises predicted upper-tail event probability by about 8--11 percentage points for annual maximum summer temperature, warm-day frequency, and warm-spell duration at 1-, 3-, and 5-year horizons. Regional bootstrap intervals remain positive for all targets and horizons. Circular-shift placebos yield one-sided probabilities of approximately 0.05--0.14 and do not survive strict family-wise correction across nine tests, so the evidence is moderate rather than decisive. Overall, annual projected memory is not a universal short-horizon predictor, but a physically interpretable inherited risk-loading variable identifying years and regions predisposed to warm-tail outcomes. --- ### GeoID-PINN: Identifiability-Aware Regional Epidemic Inference with Geographic Coupling Authors: Weixiong Hua, Fan Bu Published: 2026-07-28 Categories: cs.LG, stat.AP, stat.ML arXiv: https://arxiv.org/abs/2608.02633v1 Regional surveillance data reflect local transmission, reporting, seeding, and external infection pressure, which are difficult to identify separately. We introduce GeoID-PINN, a physics-informed neural network (PINN) for susceptible-infectious-recovered-deceased (SIRD) dynamics. The model represents spatial dependence with a row-stochastic source-composition matrix whose rows assign nonnegative source weights that sum to one. We regularize this matrix toward a spatial prior constructed from distance, adjacency, commuting, or lead-lag information. In a four-region simulation with known truth, a compatible distance prior gives source-composition error 0.099. The error rises to 0.159 without regularization and 0.577 under a strongly misspecified prior, while trajectory fit and transmission-scale estimates remain similar. Accurate trajectories therefore do not guarantee recovery of the regional dependence structure. We also evaluate GeoID-PINN retrospectively using COVID-19 data from 64 Louisiana counties. Relative to an autoregressive negative-binomial baseline, Forecast-Trained Geo-PINN reduces mean squared error (MSE) from 32,957 to 11,468 and mean absolute error (MAE) from 70.60 to 57.73. The baseline has lower negative log likelihood (NLL), 5.158 versus 5.346, indicating better distributional fit but worse point accuracy. In a controlled 15-county comparison, county adjacency reduces MSE by 6.85 percent and MAE by 3.1 percent. Similar performance across plausible priors supports structured regularization but not unique edge recovery. These results require prior-sensitivity and observation-model checks before interpretation. --- ### LLMs Can Annotate Attribution Graphs Authors: Ameen Patel, Max Zhang, Nathan Hu Published: 2026-07-28 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.02632v1 Circuit tracing is an exciting technique for revealing the internal computation of language models, but it requires a time-intensive manual step of grouping individual features or MLP neurons into supernodes. We present a simple pipeline for automating this step: directly presenting feature descriptions to a language model that groups them into supernodes. Using automated interpretability metrics, we confirm that supernodes generated by our pipeline are as interpretable as those generated by human annotators. On a two-hop Capitals task, our pipeline recovers a supernode corresponding to the intermediate hop in 97 of 100 prompts. Finally, we present a simple proof of concept using our pipeline for open-ended exploration, where we automatically annotate 1000 attribution graphs from Wikipedia prompt completions and then use an LLM judge to flag interesting graphs worth human review. We hope this work demonstrates that even simple automation can produce meaningful attribution graph annotations, motivating further work on automated circuit tracing. --- ### Noise-Robust Conditional Flow Matching: Generating Clean Samples from Noisy Datasets Authors: Adrian Urbański, Gabriel della Maggiora, Artur Yakimovich Published: 2026-07-28 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.00064v2 Generative models learn the statistical properties of their training data, so high-quality generation depends on clean and representative datasets. In scientific imaging, acquisition often yields noisy measurements, while collecting clean references can be costly, impractical or even unattainable. Training directly on these measurements results in a model that reproduces the corrupted data. This can be circumvented by learning the clean population distribution directly from the noisy data. Conditional flow matching (CFM) combines a simple regression objective with stable training, efficient sampling, and strong image-generation performance, making it a natural framework for this setting. We introduce Noise-Robust Conditional Flow Matching (NR-CFM), an unconditional generator that learns from one corrupted observation per image. NR-CFM provides a closed-form clean endpoint correction for additive white Gaussian noise and learns a data-driven correction for general Gaussian corruptions with more complex covariance structure. Across the evaluated corruption settings, NR-CFM outperforms NR-GAN in most cases and remains competitive with Ambient Diffusion in the high-noise regime. We further evaluate NR-CFM on scientific data at signal-to-noise ratios as low as $0.001$, where it generates plausible particle images from severely corrupted measurements. --- ### A Spatial Persistence Gradient in European Warming Consistent with North Atlantic Cold-Blob Influence Authors: Mauricio Herrera-Marín, Alex Godoy-Faúndez, Diego Rivera Published: 2026-07-28 Categories: physics.geo-ph, cs.LG, physics.ao-ph arXiv: https://arxiv.org/abs/2608.00063v1 Europe is warming faster than the global mean, yet the spatial organisation of this acceleration remains incompletely understood. Using ERA5 reanalysis for 1950--2024 across 28 IPCC AR6 European sub-regions, we identify two connected empirical results. First, the DFA1 Hurst exponent of interannual temperature residuals is strongly and negatively associated with the 1996--2024 warming rate ($r=-0.792$, $p=5.1\times10^{-7}$). High-persistence, mainly Atlantic-proximal regions warm more slowly, whereas low-persistence continental regions warm faster. This relationship is robust to five residualisation schemes, three memory estimators, leave-one-region-out analysis, and five null-test families, including spatial block permutation. It also persists across warming windows ($r=-0.550$ for 1981--2024, $-0.792$ for 1996--2024, and $-0.875$ for 2000--2024), but vanishes under DFA2, indicating that the signal lies in low-frequency interannual-to-decadal persistence rather than trend curvature. The pattern is consistent with, but does not prove, North Atlantic cold-blob and thermohaline influence on European land temperatures. Second, under a strict 2006--2024 holdout, contemporaneous Mediterranean SST reduces mean annual temperature RMSE by 43% (from $0.787$ to $0.449,^{\circ}\mathrm{C}$). A causal lag-weight predictor based on prior-year Mediterranean and Atlantic SST also outperforms AR(2) ($0.578$ versus $0.695,^{\circ}\mathrm{C}$) and remains informative after removing NAO, AO, and PNA effects. Similar skill from five-year moving-average and exponentially weighted predictors shows that short Mediterranean SST persistence at 1--5-year lags is the key predictive ingredient. Together, the results support a two-regime interpretation: Atlantic-proximal regions exhibit stronger memory and oceanic buffering, while continental interiors show faster warming and weaker interannual persistence. --- ### Meshy T2: Fast Native Mesh Generation with Flow Matching Authors: Jiale Xu, Rendong Liang, Yuhao Long, Siyuan Shen, Zangyueyang Xian, Xiaofei Wu, Zeyi Xu, Yuanming Hu Published: 2026-07-28 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2607.28675v3 Polygonal meshes are the standard surface representation of modern 3D pipelines, and generating high-quality meshes with artist-style topology is essential for film, gaming, and interactive 3D applications. Mainstream approaches serialize a mesh into a token sequence and decode it autoregressively, which is slow at inference and sensitive to error accumulation, making them impractical for interactive asset creation. We present Meshy T2, a fast native mesh generation framework built on flow matching. At its core is a vertex-set mesh VAE that encodes a mesh into one continuous latent token per vertex and decodes vertices, edge connectivity, and face winding order in a single pass, preserving high-precision geometry and artist-authored topology without vertex quantization or welding. Generation proceeds as a coarse-to-fine cascade of two flow-matching models: an image-conditioned voxel flow first sketches the overall shape as a coarse occupancy scaffold, and a mesh flow then populates the scaffold with per-vertex latent tokens, conditioned on the image, the scaffold, and a requested vertex budget. This design delivers three practical capabilities: interactive generation speed through parallel flow-based synthesis; effective face-count control through the requested vertex budget; and native support for multi-part assets, whose components emerge directly from the generated connectivity. In our experiments, Meshy T2 achieves state-of-the-art geometric fidelity and completes end-to-end image-to-mesh generation within a median of 6 seconds, over an order of magnitude faster than autoregressive baselines. Code and weights will be available at https://github.com/meshy-dev/meshy-t2. --- ### Meshy T2: Fast Native Mesh Generation with Flow Matching Authors: Jiale Xu, Rendong Liang, Yuhao Long, Siyuan Shen, Zangyueyang Xian, Zeyi Xu, Yuanming Hu Published: 2026-07-28 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2607.28675v1 Polygonal meshes are the standard surface representation of modern 3D pipelines, and generating high-quality meshes with artist-style topology is essential for film, gaming, and interactive 3D applications. Mainstream approaches serialize a mesh into a token sequence and decode it autoregressively, which is slow at inference and sensitive to error accumulation, making them impractical for interactive asset creation. We present Meshy T2, a fast native mesh generation framework built on flow matching. At its core is a vertex-set mesh VAE that encodes a mesh into one continuous latent token per vertex and decodes vertices, edge connectivity, and face winding order in a single pass, preserving high-precision geometry and artist-authored topology without vertex quantization or welding. Generation proceeds as a coarse-to-fine cascade of two flow-matching models: an image-conditioned voxel flow first sketches the overall shape as a coarse occupancy scaffold, and a mesh flow then populates the scaffold with per-vertex latent tokens, conditioned on the image, the scaffold, and a requested vertex budget. This design delivers three practical capabilities: interactive generation speed through parallel flow-based synthesis; effective face-count control through the requested vertex budget; and native support for multi-part assets, whose components emerge directly from the generated connectivity. In our experiments, Meshy T2 achieves state-of-the-art geometric fidelity and completes end-to-end image-to-mesh generation within a median of 6 seconds, over an order of magnitude faster than autoregressive baselines. Code and weights will be available at https://github.com/meshy-dev/meshy-t2. --- ### How Hard Does It Think? Analyzing Step-Aware Reasoning Energy in LLM Chain-of-Thought Trajectories Authors: Hui Wei, Junda Wu, Sheldon Yu, Sizhe Zhou, Yizhu Jiao, Ming Zhong, Bowen Jin, Tong Yu, Shijia Pan, Jiawei Han, Julian McAuley Published: 2026-07-28 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2607.28674v1 Understanding how computational effort is allocated across individual chain-of-thought (CoT) reasoning steps remains an open challenge: existing interpretability methods rely on output-level signals or collapse processing depth into a single trajectory-level scalar, leaving step-wise effort opaque. We propose Step-Aware Reasoning Energy (SARE), a geometric framework that quantifies effort at the granularity of individual CoT steps via Centered Kernel Alignment (CKA) between Gram matrices of token hidden states across adjacent transformer layers, capturing inter-token relational structure without requiring eigenvector alignment or cluster correspondence. SARE further contextualizes this energy within reasoning's semantic progression by modeling CoT trajectories as transitions among latent semantic states. Across six reasoning benchmarks and three open-weight LLMs, we find that reasoning energy is highly non-uniform across step types, exhibiting phase-like transitions invisible to trajectory-level metrics; incorrect trajectories show systematically lower energy at critical reasoning junctions; and SARE-based features match or outperform output-based confidence baselines in most settings, indicating that internal geometric dynamics encode predictive information beyond surface-level signals. ---