# Research Papers — 2026-09 · bundle page 95 of 266 > Full-text markdown of papers #4701–4750 of 13298 published in 2026-09, newest first. AI-readable mirrors of the full 13298-paper 2026-09 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-09/page/95 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/94.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/96.md) ### Thinking Leakage: A Causal Audit of NoThink Post-Training in Hybrid Reasoning Models Authors: Zehao Liu, Vasant G. Honavar Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28682v1 Post-training hybrid reasoning models in NoThink mode has attracted growing interest as a way to improve performance while keeping inference fast. However, these gains may draw on thinking behavior already accessible through the base model's Think mode. We formulate this thinking leakage in a causal mediation framework and audit its contribution using bidirectional interventions along a simple base-derived activation direction. Across three models and three post-training methods on competition math benchmarks, we find that leakage is real, causal, and substantial: behavioral and representational analyses reveal shifts toward Think, steering the base model along this direction reproduces most of the post-training accuracy gain, and counter-steering a checkpoint removes a substantial share of what it gains. Across nine aligned checkpoints with positive NoThink gains, the resulting leakage ratio ranges from 42% to 79%. These interventions support a substantial causal contribution of thinking leakage. Our findings show that a post-training method's apparent advantage can therefore reflect greater drift toward Think, obscuring whether it improves capability within NoThink or more effectively re-invokes existing Think behavior. --- ### Benchmarking Argumentative Behaviour of LLMs: A Study of Defences Against Character Attacks Authors: Ewelina Gajewska, Katarzyna Budzynska, Jaroslaw Chudziak Published: 2026-09-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.28673v1 Large Language Models (LLMs) are increasingly deployed as argumentative agents in persuasive dialogues, necessitating rigorous evaluation of their debating competence relative to human interlocutors. In this study, we focus on character attacks (ad hominem arguments), traditionally dismissed as fallacies, which play a pivotal role in political persuasive dialogues where ethos often rivals propositional content. Specifically, we investigate whether modern LLMs can replicate human competence to strategically use and respond to such attacks. We analyse a corpus of natural language political dialogues to identify defensive strategies human interlocutors naturally employ in ethos-centred debates and structure them into a dialogue game. Empirically, we benchmark LLM-generated dialogues against the ElecDeb60to16-fallacy corpus of U.S. presidential debates, contrasting human debaters' repertoire of defensive strategies with those of artificial agents. Results reveal a substantial difference: most LLMs rigidly prioritise logical defences, failing to exploit ethotic counterattacks as valid moves in political discourse. We argue that current safety fine-tuning constraints the strategic action space of these LLMs, making them unable to fully engage in naturalistic interactions within domains where character contestation is a normative expectation rather than a mere fallacy. --- ### Beyond Static Graph World Models: Learning Stochastic Latent Dynamics over Evolving Topologies Authors: Alex Schutz, Nick Hawes, Victor-Alexandru Darvariu Published: 2026-09-23 Categories: cs.LG, cs.AI, cs.SI arXiv: https://arxiv.org/abs/2609.28670v1 Graph-based world models have recently emerged as a means of learning transitions over relational state representations. However, existing approaches are largely limited to fixed-topology graphs or deterministic, fully observable environments. We propose the Graph Dynamics Model (GDM), a world model for graph-structured observations that is designed to handle the more general setting of evolving topologies in stochastic and partially observable environments. The GDM uses a sparse recurrent adjacency matrix to model topology updates and perform message passing, together with a recurrent state-space architecture for modelling stochastic transitions. Furthermore, we identify a gap in the evaluation of graph-based world models, as existing methods do not provide a means of comparing predicted and true distributions over the joint graph state comprising the interdependent topology, node features, and graph features. We therefore introduce the Graph Distribution Distance (GDD) metric, which uses maximum mean discrepancy with a graph kernel to comprehensively compare joint next-state distributions. We evaluate the GDM across several environments, including stochastic and partially observable settings. We demonstrate that GDM outperforms baseline models and displays zero-shot generalisation on large graphs. --- ### OPDiv: Optimal Selection of Top-K High-Scoring, Diverse Compounds Authors: Miroslav Lžičař Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28665v1 A virtual screening campaign may produce thousands of promising candidates, but only a small number can be purchased, synthesized, or tested. The practical question is how to select a set of compounds that both rank well and are diverse enough: this poses a genuine tradeoff, where selecting the highest-scoring molecules yields limited diversity, while diversity selection sacrifices some well-scoring molecules. We introduce OPDiv, a diversity selection and evaluation algorithm solving this tradeoff by finding an optimal subset of molecules using integer optimization. We demonstrate the selection algorithm in practice with fingerprint distance, shape and electrostatic diversity and compare the resulting diversity spectra. We argue that virtual screening is not merely a ranking problem, but also an implicit constrained optimization task: when redundant chemotypes are undesirable, pipelines should be compared based on the top-k compound selections satisfying the desired diversity constraints. OPDiv makes it possible to find the optimal compound set under a given diversity threshold efficiently and serves as a fair benchmark of the best diverse selection achievable by a given structure-based or ligand-based virtual screening pipeline, molecular search or generative model. --- ### Training Object Permanence in World Models Authors: Haotian Zhang, Fengyuan Yu, Dezhi Luo, Haoran Sun, Zehong Zhao, Qingying Gao, Yihan Li, Siyuan An, Huayi Qin, Yilan Zhang, Zhengze Jiang, Pinyuan Feng et al. Published: 2026-09-23 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2609.28654v1 Object permanence and solidity are hallmarks of human cognitive priors. Recent studies show that video generation models, a paradigmatic class of current world models, have begun to show emerged reasoning abilities, making them ideal candidates for building human-like physical intelligence. Do video models have emerged object permanence in them? If not, could we train them with a core-cognition inspired dataset? We introduce WROP (World Reasoning with Object Permanence), a data infrastructure of 150 hand-designed cognitive science inspired tasks, divided into six cognitive categories. We build Blender generators that randomize speed, lighting, camera angle, and other nuisance parameters while preserving each task's cognitive structure, yielding 10,000+ samples per task. We release a 1.5M-sample training corpus and a 300-question exam. On this exam we evaluate 14 video models: 3 reference-to-video, 7 edit, and 4 continuation, among which PWM-WROP, our 16B world model. In a blind pairwise Elo study, PWM-WROP ranks first among continuation models and third overall, behind only a statistical tie between two reference-to-video models. We release the data, exam, model answers, scores, weights, and PWM, our native-PyTorch training stack on AWS Trainium2. --- ### The Fellowship of the Query: Learning Retrieval Actions Authors: Mohammed Al-Maamari, Saber Zerhoudi, Michael Granitzer, Jelena Mitrović Published: 2026-09-23 Categories: cs.IR, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.28653v1 Retrieval-augmented question answering requires control decisions about when to decompose a question, search, reformulate, extract evidence, synthesize facts, verify progress, and stop. We study whether trajectory fine-tuning can improve small language models (SLMs) as next-action controllers. We additionally evaluate a low-resource setting in which a single SLM serves as both the controller and the final-answer generator. From accepted teacher search traces, we build a seven-way action-prediction task, where the model predicts the next structured teacher action from the current trajectory state, and evaluate LoRA-supervised fine-tuning across SLMs and xSLMs as controllers. On 1,646 held-out action examples, Granite 4.1 3B trained on 13,194 actions reaches macro-F1 0.6536, compared with 0.1736 for zero-shot prompting of the same model and 0.5399 for a TF-IDF logistic-regression baseline. In an end-to-end controller/generator swap evaluation over 149 held-out trajectories, using the fine-tuned model for both roles improves Exact Match from 0.7530 to 0.7946 and token F1 from 0.7783 to 0.8295 compared with using the base model as both controller and generator. The cross-role conditions show that the fine-tuned controller increases evidence-fact recording when the generator is fixed, while controller-only final-answer gains are not statistically clear. Overall, trajectory supervision improves action prediction and evidence-recording behaviour in this evaluated pipeline. Code is available at https://github.com/padas-lab-de/agent-action-controller --- ### PePESeg3D: Perception Prior Enhances Multi-Scale Segmentation for 3D Gaussian Splatting Authors: Sungjae Choi, Seunghee Koh, Junmo Kim Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28645v1 Recent advancements in 3D Gaussian Splatting (3DGS) have extended its capabilities to multi-scale segmentation. Existing methods reconstruct a scene with Gaussian primitives and learn multi-scale segmentation features separately, which leaves the geometry unaware of semantic structure and the feature learning dependent on incomplete mask supervision. To address these limitations, we present PePESeg3D, a novel framework that injects perception priors into a multi-scale 3D Gaussian segmentation pipeline. To fully exploit perception priors, we integrate them not only into contrastive feature learning but also into the upstream geometry reconstruction. Specifically, PePE Reconstruction incorporates monocular depth and mask constraints to ensure semantically coherent object structures. Building on this aligned geometry, PePE Contrastive Learning leverages dense depth-color cues and view-consistent centroid supervision to compensate for the incompleteness of multi-scale masks obtained from a 2D foundation model. Extensive experiments on the SPIn-NeRF, LERF-Mask, and NVOS benchmarks demonstrate that PePESeg3D achieves state-of-the-art performance in both multi-scale segmentation and scene reconstruction, highlighting the importance of integrating perception priors into both geometry optimization and feature learning for accurate multi-scale 3D segmentation. Our code is available at https://github.com/BeCow5X5/PePESeg3D. --- ### RLVR landscapes for iterated multiplications can be benign: Insights from spin-glass theory Authors: Noa Rubin, Zohar Ringel Published: 2026-09-23 Categories: cs.LG, cond-mat.dis-nn, stat.ML arXiv: https://arxiv.org/abs/2609.28625v1 Despite the importance of reinforcement learning with verifiable rewards (RLVR), the extent to which it can learn new reasoning capabilities remains debated. Here we study the optimization landscape of RLVR on algorithmic tasks, such as iterated group and quasigroup multiplication. To this end, we map entropy-regularized RLVR over myopic tabular policies onto an energy-based (spin-glass) model over deterministic policies. This mapping upper-bounds what RLVR can achieve, and lets us rigorously characterize the landscape in this tabular setting. We show, both theoretically and experimentally, that for a wide class of models and tasks with uncorrelated inputs, this landscape is benign, containing no local minima that could trap RLVR training. Rather, the practical difficulty of these tasks appears to stem, at least in part, from issues such as diffusive barriers and gradient-estimation error in traversing the landscape. These are genuine obstacles that can prevent a solution from being found, but they are distinct from the landscape itself being rugged. We show that these obstacles can often be mitigated through the choice of entropy regulator. Consistent with this theory, we find that a transformer trained from scratch, using only last-token rewards, successfully learns an algorithmic chain of thought for iterated non-Abelian group multiplications. --- ### Reward Hacking Challenges Oversight of Autonomous Research Agents Authors: Yue Huang, Zhangchen Xu, Yuchen Ma, Wenjie Wang, Zheyuan Liu, Ziwei Xu, Pin-Yu Chen, Michel Galley, Zinan Lin, Stefan Feuerriegel, Radha Poovendran, Misha Sra et al. Published: 2026-09-23 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.28614v1 Autonomous research agents can design experiments, evaluate results, and write reports, giving them control over both a scientific result and the evidence used to support it. This creates a risk of reward hacking: meeting the reward criteria without achieving the intended goal. We study (1) how often models reward-hack without instructions to do so, (2) how effective and detectable their methods are when hacking is allowed, and (3) how they adapt when an LLM review panel returns its decision and reasons. Across 17 language models and 38 tasks, the spontaneous reward-hacking rate is 30.5% on open-ended research-pipeline tasks and 2.9% on task-specific kernels. When hacking is allowed on tasks whose pass thresholds exceed our best compliant baselines, 505/677 attempts (74.6%) are confirmed reward hacks: they both clear the threshold and receive mechanism-verification panel confirmation of an evaluation exploit. An LLM panel reviewing only submitted code and reported scores misses 33/505 confirmed hacks (6.5%). Direct methods that achieve the highest scores are often easy to detect, while less direct methods evade more often. In a five-round loop, the number of model-task pairs with an evasion rises from 7 to 56. Among 79 pairs evaluated under two feedback conditions, cumulative evasion reaches 40.5% with detailed feedback and 20.3% with generic rejection. The detailed condition includes the review decision, reasons, and attempt history, so this comparison does not isolate the effect of explanations. These findings highlight the need for stronger defenses, including metrics kept outside the agent's control and independent recomputation on data chosen to expose likely exploits. --- ### Decision Hijacking: Prompt Injection Attacks on Jev's Typed Probabilistic Decisions Authors: Tiantong Wu, Wei Yang Bryan Lim Published: 2026-09-23 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2609.28613v1 Most studies of prompt injection focus on generative agents, leaving their effects on models with schema-defined outputs unclear. We examine these effects in Jev, a non-generative decision model, using 510 reconstructed InjecAgent cases. Malicious content shifts action probabilities but rarely causes Jev to select the attacker's target. Override markers reduce this influence, while claims of contextual relatedness have small effects. Adaptive attacks using score feedback double the mean highest attacker-target probability found during optimization, while success on fresh validation calls rises from 1.8% to 3.5%. Exploratory analysis links these successes to small initial decision margins or greater attacker control over the observation. Together, these findings show that schema-defined outputs change but do not eliminate prompt-injection risk, highlighting the need to evaluate how untrusted content influences choices within the allowed action set. --- ### Global Convergence of Third-Order Langevin Dynamics for Non-Convex Optimization via Simulated Annealing Authors: Yingli Wang, Kelvin Shuangjian Zhang, Lingjiong Zhu Published: 2026-09-23 Categories: math.NA, math.OC, math.PR, stat.ML arXiv: https://arxiv.org/abs/2609.28611v1 We study global convergence guarantees of third-order Langevin dynamics for non-convex optimization via simulated annealing with fixed friction and decreasing noise. An explicit three-block distorted entropy transfers dissipation from the noisy auxiliary variable to the full state. Under dissipativity, regularity, and low-temperature functional-inequality assumptions, logarithmic cooling drives the objective values to the global minimum in probability at the barrier-controlled kinetic rate. For the exact-force-integral and midpoint three-stage discretizations, polynomially decreasing steps preserve this rate on the physical time scale. The cubic local endpoint estimate gives a less restrictive sufficient step-size condition than the available frozen-force kinetic result. A comparison with the one-gradient UBU integrator shows how its centered stochastic local error leads, under the same strong-coupling analysis, to a smaller sufficient iteration exponent. Numerical experiments are conducted to illustrate our theory. For a double well objective, third-order Langevin terminal-success point estimates are higher than UBU at both a common horizon and an equal gradient budget. For a high-dimensional nonconvex neural-network objective using synthetic data, independently tuned UBU and third-order Langevin schemes both outperform overdamped Langevin dynamics; the third-order Langevin point estimate is higher. For the same neural-network objective on real data, we show the same point-estimate ordering for best-basin probability and post-quench test accuracy. Numerical code and associated experiment results are publicly available at https://github.com/gagawjbytw/simulated-annealing-third-order-langevin. --- ### UltraBench 2: Towards Robust Evaluation of Vision Foundation Models on Ultrasound Authors: Ashwath Radhachandran, Adam Tupper, Christian Gagné, William Speier Published: 2026-09-23 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.28610v1 Benchmarking is an increasingly critical part of research in machine learning and the domains where it is applied, including healthcare. Yet, despite the steady development of new ultrasound foundation models in recent years, the development of well-designed benchmarks to evaluate them has lagged behind. This deficiency has led to fragmented and inconsistent evaluations of competing models, making it difficult to measure progress. To address this issue, we introduce UltraBench 2, a comprehensive benchmark with wide anatomical and task coverage, and a focus on standardization, reproducibility, and ease-of-use. Using this benchmark, we compare existing vision foundation models for ultrasound image analysis. Our analyses demonstrate that ultrasound-specific pretraining still leads on classification, but that state-of-the-art general-purpose models have drawn level on segmentation. --- ### Adversarial Closed-Loop Curriculum for Evolving Role-Playing Agents Authors: Zheng Zhang, Liu Liu, Qi Chai, Deheng Ye, Peilin Zhao, Mao Zheng, Hao Wang Published: 2026-09-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.28609v1 Role-playing agents based on large language models have been widely applied in areas such as personalized assistance and social simulation. Recent RL methods typically train on a fixed scenario pool collected before learning begins. This creates a distributional bottleneck: as the agent improves, the scenarios where it performs poorly also change, while the training distribution remains static. Therefore, we propose AdvRole, an adversarial context rewriting framework that turns role-playing RL into a closed-loop curriculum. AdvRole alternates between an Actor that learns to role-play and a Rewriter that edits character profiles and dialogue contexts into actor-specific hard scenarios. The Rewriter is trained with a performance-gap reward, which favors rewrites that reduce the current Actor's score relative to the original scenario. As a result, the scenario pool evolves with the Actor and continuously targets under-mastered regions of the character-context space. Experiments on three role-playing benchmarks covering English and Chinese, as well as a new multilingual benchmark we release, show that AdvRole consistently outperforms baselines. --- ### fable.intermittent: benchmarking probabilistic forecasting methods for intermittent time series Authors: Stefano Damato, Lorenzo Zambon, Giorgio Corani, Dario Azzimonti Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28607v1 Intermittent time series are common in spare-parts demand and retail sales. Since the cost of forecast errors is typically asymmetric, decisions such as inventory control require the full predictive distribution rather than a point forecast. Many probabilistic forecasting methods have been proposed; their implementations, however, are scattered across different software frameworks, making it difficult to compare them systematically. We introduce fable.intermittent, an R package that implements several probabilistic forecasting methods for intermittent series within the fable framework. The package allows several models to be fitted and evaluated on a collection of time series through a single, simple forecasting pipeline. We also introduce TWEES, a new exponential smoothing model with a Tweedie predictive distribution. Fitting TWEES requires repeated evaluation of the computationally demanding Tweedie density. We also release the R package tweedieDistr, whose implementation of the Tweedie distribution is substantially faster than the existing one while preserving the same numerical accuracy. We evaluate the methods implemented in fable.intermittent on four datasets, also released in the package. --- ### UO-FIE: Combining Exact-Label Supervision with Graded Utility for Factivity Inference Authors: Xinchen Xiao Published: 2026-09-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.28605v1 The Factivity Inference Evaluation 2026 (FIE2026) classifies Chinese context-hypothesis pairs into nine ordered factivity intervals. Its evaluation metric rewards both exact predictions and proximity to the correct interval, while 64.1% of the 566 training examples belong to a single class. In preliminary experiments, several mDeBERTa classification models predominantly predict the dominant class, whereas a Huber-regression baseline produces more predictions near the correct interval but fewer exact matches. We introduce Utility-Oriented Factivity Inference (UO-FIE), a parameter-efficient system that combines exact-label supervision with graded utility. UO-FIE predicts a distribution over the nine classes and combines hard-label supervision, utility-based soft targets, scheduled class weights, and an ordinal loss. We evaluate expected-utility decoding in controlled comparisons and use ordinal calibration selected on out-of-fold predictions for the submitted system. Based on Qwen3.5-9B with LoRA, UO-FIE ranks first in the fine-tuning track with a macro utility of 0.8316. A separate prompt-based ensemble ranks third in the non-fine-tuning track with a macro utility of 0.8450. --- ### Physics-Informed Self-Supervised Learning for Joint Wire Calibration and Interaction Position Reconstruction in Multi-Wire Parallel Plate Avalanche Counters Authors: Antoine Lemasson, Maurycy Rejmund Published: 2026-09-23 Categories: cs.LG, nucl-ex, physics.ins-det arXiv: https://arxiv.org/abs/2609.28604v1 Scientific instruments require accurate calibration to convert detector signals into reliable physical observables. Conventional calibration procedures typically rely on dedicated calibration measurements, analytical response models or labelled reference data, limiting their ability to adapt to changing operating conditions and detector aging. We present a physics-informed self-supervised learning framework that jointly performs wire calibration and interaction position reconstruction in Multi-Wire Parallel Plate Avalanche Counters (MWPPACs) without requiring labelled position measurements or dedicated calibration runs. The method formulates detector calibration as a latent optimization problem in which global wire gains and event-wise interaction positions are estimated simultaneously using supervision derived exclusively from detector geometry and charge-energy consistency constraints. A detector-independent neural network reconstructs sub-wire interaction positions from local charge distributions, eliminating the need to assume analytical induction profiles by learning the detector response directly from experimental data. The end-to-end differentiable framework enables continuous detector self-calibration while improving the uniformity and accuracy of position reconstruction. Experimental evaluation on the entrance MWPPAC tracking detectors of the VAMOS++ magnetic spectrometer demonstrates stable convergence, improved spatial homogeneity and enhanced position resolution. Beyond the detector studied, the method establishes a general framework for physics-informed self-supervised calibration of scientific instruments and is a step toward autonomous intelligent instrumentation capable of continuous adaptation during operation. In this paradigm, detector calibration is no longer a prerequisite for an experiment but an integral part of the measurement process itself. --- ### Learning to Discover Interesting Mathematics Authors: Niket Patel, Ahmad Rammal, Amaury Hayat, Remi Munos, Julia Kempe Published: 2026-09-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.28603v1 Recently, Large Language Models (LLMs) have been increasingly able to solve advanced mathematical problems, including many that have been open for decades. This opens the door to expansion of mathematical knowledge at unprecedented scale. Yet, while LLMs may be able to conjecture and prove more and more theorems, it remains open whether this new mathematical knowledge is interesting or useful. We define intrinsic interestingness of a theorem as the ratio between the length of its proof and the length of its statement. We show that this correlates strongly with an extrinsic measure of the downstream utility of a theorem. We identify the difficulty of a proof conditioned on a set of premises as a useful primitive for computing these metrics, and train a 27B model that predicts proof difficulty more accurately than frontier general-purpose models. Optimizing for our metric creates a model capable of producing more interesting theorems, while also reducing substantial or full overlap with Mathlib from 91.9% to 30.6%, showcasing the creation of more out-of-distribution math. We show that our system can generate candidate theorems, select the most interesting among them, and iteratively build on a self-expanding mathematical library. These metrics provide a practical and quantifiable signal for ranking conjectures and guiding proof search within formal mathematical libraries. Our framework provides a path towards self-expanding, machine-verified mathematical libraries that can choose worthwhile statements without relying on human-supplied targets. --- ### HClimRep-Ocean: A Global Ocean Emulator on an Unstructured Mesh Authors: Kacper Nowak, Aleksei Koldunov, Nikolay Koldunov, Savvas Melidonis, Ankit Patnala, Simon Grasse, Julius Polz, Christian Lessig, Martin Schultz, Thomas Jung Published: 2026-09-23 Categories: physics.ao-ph, cs.LG arXiv: https://arxiv.org/abs/2609.28601v2 Machine-learning (ML) emulators for atmospheric processes have advanced rapidly in recent years, transforming weather forecasting. Although early ML ocean forecasting models now exist, they remain less developed than their atmospheric counterparts. Unlike the atmosphere, much of the ocean's kinetic energy resides in mesoscale eddies whose characteristic spatial scales are approximately an order of magnitude smaller than those of comparable atmospheric features. Moreover, complex coastlines, narrow straits, and ice-covered seas make boundary representation a central challenge that atmospheric models do not face. Consequently, numerical ocean simulations commonly use locally refined or even completely unstructured meshes. However, their data-driven counterparts have so far been built around latitude-longitude grids. We present HClimRep-Ocean, an ocean emulator that operates directly on the native unstructured mesh of FESOM2. The emulator is trained on a 209-year AWI-CM3 control integration and is run without atmospheric forcing, receiving the atmospheric state only at initialisation time, which isolates the predictability carried by the ocean state itself. Skill is strongly field-dependent: for currents, HClimRep-Ocean outperforms every reference at 30 day forecast, whereas for temperature and salinity a damped-anomaly persistence forecast remains the more accurate estimator. This behaviour is physically interpretable: current variability is largely geostrophic and internally generated, whereas sea-surface temperature and salinity fluctuations are driven by atmospheric forcing through weather state. Evaluated independently on the OceanBench benchmark, a reanalysis-trained variant of HClimRep-Ocean achieves the lowest RMSE against GLORYS reanalysis among all assessed systems, confirming the competitiveness of the native-mesh approach. --- ### BRFID: Toward Byzantine-Robust Federated Intrusion Detection Authors: Asmah Muallem, Firdous Kausar, Sajid Hussain, Lei Qian Published: 2026-09-23 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.28599v1 Flipping 60\% of training labels from a single Byzantine client using label-flipping model poisoning self-degrades an attacker's own federated detection accuracy, $99.96\%$ (at no poisoning rate) to $84.33\%$ in a three-client federated IDS. Where the Federated global ensemble maintains stable accuracy across all tested poison rates, without a defense mechanism in place and without coordination between attackers. In this paper, we present empirical results quantifying the impact of label-flipping poisoning attacks on a three-client federated IDS trained on CICIDS2017 with non-IID attack subtype distributions across clients. We demonstrate that the signal of the adversarial self-compromise represents a detectable anomaly for exploitation for Byzantine client identification in the absence of target data exfiltration. We note that the aggregation step uses a Federated Forest (tree concatenation) rather than a parametric FedAvg; the results therefore measure the impact of poisoning on per-client performance under ensemble aggregation, and extension to genuine FedAvg with a parametric classifier is planned for future work. --- ### TAM-Chain: Multi-Scale Thyroid Cytology Classification via Absorbing Markov Chains and Shannon Entropy Uncertainty Quantification for False-Negative Suppression and Domain-Shift Adaptation Authors: Hai Pham Ngoc Published: 2026-09-23 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.28590v1 Background & Problem: Thyroid Fine-Needle Aspiration Biopsy (FNAB) cytology based on the Bethesda System plays a pivotal role in early thyroid cancer detection; however, deep learning approaches face substantial challenges regarding high false-negative rates and overconfidence under clinical domain shift. Methods: In this study, we propose TAM-Chain, a multi-scale (10x, 20x, 40x) thyroid cytology classification framework leveraging Absorbing Markov Chain theory combined with Shannon Entropy-based Uncertainty Quantification. The framework dynamically models multi-magnification feature extraction as an absorbing stochastic process, enabling optimal stopping criteria and a human-in-the-loop referral mechanism to strictly suppress critical diagnostic errors. Results: Extensive evaluation on an internal test set (N = 235) demonstrates a Macro F1 score of 0.9741 with an absolute False-Negative Rate (FNR) of 0.00%. On an independent external validation set (N = 1015) presenting severe domain shift, TAM-Chain maintains superior stability and classification performance (Macro F1 = 0.7026) by adaptively adjusting the expected stopping step and triggering specialist referrals, significantly outperforming single-magnification baselines. Conclusion: The TAM-Chain framework proves to be a highly effective, safe, and adaptable solution for digital pathology workflows, successfully harmonizing automated diagnostic efficiency with stringent biological safety. --- ### NumericJev: Jev-like LLM Numerical Decoding with Multiway Decision Trees Authors: Weiwei Ye, Hangchen Liu, Renhe Jiang Published: 2026-09-23 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.28587v1 Large language models can interpret natural lan- guage, yet robust decisions remain challenging. Jev-like models expose structured choices, but these interfaces do not directly provide numeri- cal values at a requested precision. We propose NUMERICJEV, a training-free numerical decod- ing algorithm that enables numerical output from any LLM with a Jev-like structured-choice in- terface. Surprisingly, on our arithmetic bench- mark, it outperforms direct selection from a can- didate list containing the correct answer by 2.93 percentage points (Figure 1). Our motivation comes from the observation that numerical range selection is itself a decision problem that Jev- like LLMs can address. NUMERICJEV recur- sively refines a range through a multiway deci- sion tree while retaining the original question in context, without parameter updates or hidden- state access. On a 100-value grid, a ten-way tree requires only two decision rounds. Range- normalized MAE is 1.84% versus 5.18% for di- rect choice. A separate three-date historical- index study yields 4.58% mean relative recall er- ror and 0% readout error when the value is sup- plied. Code is available at https://github. com/Bring-AI/jev-numeric. --- ### Persistent Billable State: Denial-of-Wallet Attacks and Defenses in Tool-Calling LLM Agents Authors: Jinqian Zhang, Haojun Xia, Shujiang Wu, Jingkun Yue, Xia Zhang, Zhangpei Cheng, Bibo Tu Published: 2026-09-23 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2609.28585v1 Multi-step tool-calling LLM agents rely on host runtimes to preserve state across turns. When a runtime carries an external tool return into later model inputs, providers meter it again. An admitted malicious or compromised tool can thereby convert untrusted data into recurring victim-billed processing without victim credentials or local runtime privilege. We call retained content persistent billable state and formalize the host's decision over whether and how it enters later billable context as the persistent billable-state boundary. We present the first systematic security study of this post-admission lifecycle. We derive six denial-of-wallet attack vectors and build DOW-BENCH, an end-to-end harness evaluated across six model families. Across 243 executions, usage telemetry shows that the maximum per-session cumulative input reaches 14,293x the session's first-call input. Controlled history-policy reruns isolate raw retention's contribution: retaining raw history increases mean effective session cost by 21.2-35.9%. Compression succeeds on 10/12 and 11/12 history-dependent tasks, versus 2/12 under deletion for each provider. To govern this boundary, we combine deterministic history transformation with four host-side invariants that bound prompt mass, context growth, recursive opportunity, and cumulative spend before reingestion. The kernel contains every recurring attack in the 123-evaluation replay corpus. Across 24 Mistral Small 4 workflows, a progress-authorized policy achieves 22/24 oracle-verified task successes with no pre-completion interruptions, versus 13/24 under a fixed cap. Only 71 of 3,830 scanned MCP server and transport repositories expose any code-visible safeguard proxy, and none cover all four safeguard families. These results establish persistent billable state as a first-class security object and pre-reingestion as its host-owned control point. --- ### SGA: Uncertainty Quantification for Multi-Step Forecasting in Time Series Foundation Models Authors: Xin-Yu Hu, Shuang Liang, Cheng Feng, Shao-Qun Zhang Published: 2026-09-23 Categories: cs.LG, cs.AI, stat.ML arXiv: https://arxiv.org/abs/2609.28582v1 The recent emergence of Time Series Foundation Models (TSFMs) has significantly advanced multi-step forecasting performance, enabling accurate predictions over extended future horizons. However, existing TSFMs often suffer from significantly inherent uncertainty, which typically manifests as derived forecast branches emerging at each time step and spreading to subsequent steps; different forecast branches often exhibit varying forecasting performance, thereby undermining the credibility of TSFM forecasts. In this paper, we propose the Slicing-Graphing-Alignment (SGA) method to quantify the uncertainty of multi-step TSFM forecasts. The proposed SGA first characterizes the topology of all potential forecast branches using a directed acyclic graph, such that the graph complexity bounds the uncertainty of multi-step forecasts, and then precisely measures the graph complexity by integrating both topological information and TSFM-inherent stochasticity. Experimental results conducted on 11 TSFMs and 27 datasets demonstrate that (i) SGA achieves the best performance when ranking predictive errors with uncertainty estimates; (ii) SGA works with a more extensive and more precise sampling coverage than those of existing UQ methods, deriving a quantification mechanism fundamentally different from those of established ones; and (iii) larger model scales of TSFMs correlate with lower uncertainty estimates of multi-step forecasts, suggesting another empirical scaling law for uncertainty quantification of multi-step TSFM forecasts. --- ### Auditability Is Not One Property: Rule Overlap, Behavioural Agreement, and Composition in Reinforcement Learning Authors: Liu Hung Ming Published: 2026-09-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.28581v1 Reinforcement-learning (RL) policies are often distributed as opaque neural checkpoints, while training logs show that a run occurred without explaining what the policy learned. We study whether independently trained policies can be represented and composed through auditable discrete behavioral rules. We define auditability as six separately testable predicates: trace integrity, lossless coding, rule coverage, behavioral agreement, composition quality, and value-model reliability. Our protocol uses a shared frozen symbolizer, passive rule extraction, an append-only hash-bound ledger, exact environment replay, and offline confidence-ranked arbitration with an explicit blind-spot fallback. The results place strict limits on this description layer. Rule-set overlap does not imply behavioral agreement: policies may share symbolic rules while choosing near-chance-matching actions on fresh states. The fused policy therefore selects among existing rules rather than generating a new skill. On a conflict-dominated task, an apparent fusion failure is traced to an induction/deployment mismatch: rules induced from sampled actions were evaluated under argmax actions, and deployment-consistent re-induction reverses the arbitration ordering. A fitted-Q generalized-policy-improvement diagnostic also fails in both environments, limiting claims that rule fusion is superior to value-based composition. One exploratory comparison favors rule fusion, but its comparator is post hoc, the task is partly saturated, and the fused policy remains below the strongest held-out actor. We contribute an evidence-bounded audit and composition protocol, not a claim of universal interpretability or autonomous skill generation. Future work must add temporally extended skills, cross-skill interfaces, composition search, and independent novelty audits. --- ### Token Clustering and Semantic Sequence Mamba for Hyperspectral Image Classification Authors: Yimin Zhu, Mahmood Elahi, Lincoln Linlin Xu Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28580v1 Although hyperspectral images (HSIs) provide rich spectral-spatial information, accurate pixel-level classification remains challenging because of spectral-spatial heterogeneity and complex spatial structures. Existing vision state space models (Mamba) typically construct sequences according to predefined spatial neighborhoods, without explicitly accounting for semantic similarity or spatial non-stationarity. To address this limitation, we propose Token Clustering and Semantic Sequence Mamba (STMamba), which organizes sparse tokens into semantically coherent sequences for hyperspectral image classification with the following features. First, at the macro level, a hierarchical encoder decoder progressively selects semantic tokens with the Token Clustering Module (TCM) and restores dense features using a parameter-free Cross-scale Neighborhood Attention (CNA) Upsampler. Second, at the micro level, TCM first identifies representative cluster centers through density-aware clustering and estimates soft memberships based on feature similarity. A quadtree-based dynamic selection strategy then retains sparse and spatially distributed tokens from each semantic cluster, forming coherent semantic-token sequences while reducing redundant pixel-wise representations. Third, parallel Spatial and Spectral Semantic-wise Sequencing Mamba (SWSM) modules capture complementary long-range spatial and spectral dependencies within homogeneous semantic token sequences while suppressing irrelevant interactions across heterogeneous regions. Experimental results on three large-scale benchmark datasets demonstrate that STMamba outperforms the SOTA methods with respect to quantitative and qualitative results. --- ### Uncovering Residential PV-EV Co-Adoption from Smart-Meter Data: Load Archetypes and Detection for Demand-Side Planning Authors: Jack Zheng, Hao Wang Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28578v1 The increasing adoption of electric vehicles (EVs) and rooftop photovoltaic (PV) systems is reshaping residential electricity demand and creating new challenges for demand-side management (DSM), tariff design, and low-voltage network planning. Much of the existing literature examines EV charging or PV generation in isolation, leaving the behavioral dynamics of household co-adoption less understood. We develop an integrated, two-part workflow to analyze advanced metering infrastructure (AMI) data. A discovery component applies dynamic time warping (DTW) k-means with DTW barycenter averaging to cluster daily import or export profiles into interpretable behavioral archetypes, while a predictive component trains a bidirectional long short-term memory (BiLSTM) model on 21-day windows and benchmarks it against tabular baselines for PV/EV activity detection. The EV activity labels are inferred from charging-like load signatures because charger measurements are unavailable. Using half-hourly AusNet residential data from Victoria, Australia, the clustering uncovers distinct patterns across PV-only, EV-only, co-adoption, and neither cohorts; for co-adopters, a midday-centered weekday export archetype accounts for approximately 50% of days. At validation-tuned thresholds, both BiLSTM and XGBoost achieve strong discrimination. BiLSTM obtains 0.991 for the area under the receiver operating characteristic curve (AUROC), 0.906 for macro-F1, and the highest recall on the most difficult class (0.836 for EV-only recall). Tree-based baselines remain competitive. Performance remains stable across plausible labeling rules (macro-F1: 0.894--0.914) and strictly forward temporal splits (macro-F1: 0.894--0.906). --- ### Time-Series Foundation Models That Understand Data Revisions Authors: Taimoor Ahmad Published: 2026-09-23 Categories: cs.LG, cs.SE arXiv: https://arxiv.org/abs/2609.28576v1 Historical observations are not always fixed: statistical agencies revise previously published values as new evidence arrives. Forecasting from a contemporary download can therefore expose a model to information unavailable at the date it purportedly made a prediction. We propose VINTAGE-TS, a revision-aware adaptation of a time-series foundation model that distinguishes observation time from information-availability time. Its targets are the next period's first-published value and the value available a fixed number of days after that publication; neither is declared final truth. A joint predictive distribution preserves dependence between these targets and exposes uncertainty about their difference. We specify an ALFRED-based rolling evaluation, a matched Chronos-2 comparison, conventional and revision-aware baselines, and a separate audit of pretraining overlap. The accompanying software implements validity-interval reconstruction, delayed-label filtering, a frozen-backbone adapter interface, and reproducible diagnostics. An executed synthetic demonstration and a 25-configuration sensitivity suite verify the workflow, expose variation across seeds and revision regimes, and illustrate how hindsight contamination changes measured performance. Thirty one automated tests check temporal and integration contracts. Real ALFRED and Chronos-2 experiments have not been executed; no empirical foundation-model advantage is claimed. --- ### TWIST: A Proposed Benchmark for Intervention Quality in Conversational Memory, with a Human-Validated Draft-Alignment Authors: Subrat Panda Published: 2026-09-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.28575v1 Long-conversation memory benchmarks increasingly test recall and prompted knowledge updates, and recent work studies evolving user beliefs and memory state. TWIST is a proposed benchmark suite for a complementary, unmeasured property: intervention quality -- whether a deployed memory system, exercised through its own ingest/recall/vet surface, acts correctly at belief change points. Four tracks cover unprompted tension detection, vetting outgoing drafts against the record, answering with current beliefs while preserving supersession history, and governing sensitive recall. The suite extends LoCoMo's corpora and harness, pairing every detect/block metric with a matched do-not-over-detect control: surface-matched hard negatives price false intervention, so no track can be gamed by flagging everything. The benchmark itself is validated first: independent, gold-blind double annotation with adjudication, judge decoy calibration, and a separability audit. On the human-validated Track B v1.0 key (161 items, post-adjudication kappa = 0.85), no tested configuration simultaneously achieves high contradiction recall, high hard-negative specificity, and high attribution: flat-RAG baselines detect 0.76-0.97 of true contradictions but falsely flag 16-43% of surface-matched safe drafts depending on backend, while a deployed coherence-oriented system almost never over-flags (0.98-1.00 specificity) yet catches 42% of true contradictions -- a trade-off no recall-only score can see. A 13-configuration baseline ladder localizes causes: every gold contradiction is detectable from its evidence alone (recall 1.000), calibrated models nearly solve the track given the full transcript -- consistent with substantial retrieval-coverage gaps -- and draft-only floors reveal model-dependent style priors. A system's TWIST profile, beside its recall score, measures whether memory knows when to intervene and when not to. --- ### Where Cyber Agents Struggle: Bottleneck Analysis of Multi-Stage LLM Agents Authors: Saeedeh Lohrasbi, Mohammad Mamun, Ahmed Yehia, Scott Buffett, Sherif Saad Published: 2026-09-23 Categories: cs.CR, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2609.28572v1 Multi-stage LLM-based cyber agents may complete attack workflows while remaining brittle, costly, or reliant on incorrect interpretations of execution evidence. Success rates alone obscure inefficiency, adaptation through retries, and recognition of success or failure. We present an end-to-end diagnostic study of an Autonomous Adversary system with orchestrator, executor, and validator LLMs in enterprise-like lateral-movement scenarios. Six frontier models are evaluated across two scenarios and three modes: expert-defined, self-scaffolded, and fully autonomous. We assess validator consistency and evidence grounding; introduce a subtask-conditioned, cost-aware score for abnormal token use, retries, and runtime; and use comparative LLM-as-a-Judge analysis to identify planning deficiencies, including tool misalignment, plan similarity, over-specification, inadequate probing, and weak recovery. Validators are generally relevant and evidence-grounded but often nonspecific and overly optimistic. Bottlenecks cluster in credential and lateral-movement tasks, spread with scenario complexity, and vary more under full autonomy. Reliable evaluation must assess outcomes, evidence interpretation, resource use, and adaptation after failure. --- ### DEEPO: Dual-Entropy Enhanced Policy Optimization for Hallucination in MLLMs Authors: Yingxuan Zhuang, Miao Pan, Wangjie Gan, Jingxiao Yang, Fan Wang, Weiming Liu, Cheng Tan, Xuhong Zhang, Jintao Chen Published: 2026-09-23 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.28570v1 Reinforcement learning (RL) is widely used to sharpen reasoning in multimodal large language models (MLLMs), yet its effect on hallucination is uneven. We trace this to two weak points in the \emph{correction chain} from reward to parameter update. At the rollout level, hard queries---those with high semantic entropy---frequently produce unanimously wrong sample groups, collapsing the group-relative advantage to zero exactly where hallucination risk is highest. At the optimization level, confident-but-wrong tokens are gradient-invisible: a categorical policy's expected score-gradient norm vanishes as its distribution sharpens, so the predictions that most need correction receive the weakest updates. We propose Dual-Entropy Enhanced Policy Optimization (DEEPO), a dual-stage enhancement combining signal variance regularization with gradient preconditioning: semantic-entropy-triggered expert prefixes inject grounded continuations on high-uncertainty queries, providing direct supervision and restoring advantage variance, while advantage-sign-aware Renyi preconditioning counteracts logit-level saturation so correction reaches confident errors in the operational confidence regime. Both branches improve over GRPO individually; their interaction is statistically significant on VideoMMMU---the most complex long-horizon task in our evaluation suite (+4.0$, 95\% CI (1.1, 6.9))---and additive elsewhere. DEEPO reduces hallucination while preserving accuracy and training stability. --- ### Leakage-Safe Machine Learning for Hydrogen Embrittlement Detection in 316L Stainless Steel: A Region-Held-Out Evaluation of Texture and Deep Features in SEM Micrographs Authors: Muhammad Awais, Muhammad Yaseen, Abdul Shakoor, Niaz Ahmed Niaz, Huria Zia, Muhammad Zain Shakoor Published: 2026-09-23 Categories: cs.LG, cond-mat.mtrl-sci arXiv: https://arxiv.org/abs/2609.28567v1 Scanning electron microscopy (SEM) is routinely used to characterize the microstructural changes caused by hydrogen embrittlement (HE) in structural steels. Machine learning can automate this characterization, but models are often evaluated using image-level splits. When several images come from the same specimen region, such splits leak information between the training and test sets. Here, we propose a region-held-out protocol for classifying as-received (AR) and hydrogen-charged (H2) SEM micrographs of 316L stainless steel, based on Leave-One-Region-Out (LORO) cross-validation over 14 spatial regions (8 AR, 6 H2; 31 images). We compared six feature-classifier combinations built on local binary patterns (LBP), grey-level co-occurrence matrices (GLCM), self-supervised convolutional embeddings pretrained on 143 unlabeled SEM images, and a convolutional neural network (CNN). The simplest texture approach, LBP with a support vector machine (LBP+SVM), performed best, achieving a balanced accuracy of 0.79, H2 recall of 0.69, and H2 precision of 0.82, outperforming every deep-learning and combined-feature model. A group-level permutation test (500 permutations sampled from the 3,003 possible region-to-label assignments) yielded p = 0.008, indicating that the result cannot be explained by a chance alignment of the region structure. Grad-CAM maps from a CNN trained on the full dataset tended to concentrate on localized surface and grain-boundary features, where hydrogen-induced morphological changes are known to occur. Under a leakage-safe, statistically validated protocol, texture descriptors recover a hydrogen-charging signature from SEM micrographs even with few samples, and the same protocol can be extended to larger HE detection studies in other alloy systems. --- ### When Explanations Cannot Be Read: Measuring and Correcting SHAP and LIME Rendering for Right-to-Left Languages Authors: Rameesha Zia, Muhammad Shahid Iqbal Malik Published: 2026-09-23 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.28565v1 Post hoc explanation methods such as SHAP and LIME are widely used to interpret text classifiers, but their visualizations are mainly designed for left-to-right languages. When applied to right-to-left (RTL) languages such as Urdu, Arabic, Persian, and Hebrew, the attribution values remain mathematically valid, while their visual presentation fails. Tokens appear out of sequence, connected letterforms break apart, and plot layouts do not follow the natural reading direction. This study addresses this gap as a visualization problem rather than a limitation of the explanation methods themselves. We present SHAP-RTL, a rendering layer that corrects reading direction and script shaping in SHAP and LIME visualizations, with per-language font selection, while preserving the original attribution values, feature ordering, and model outputs. The approach is evaluated on Urdu, Arabic, Hebrew, and Persian hate and offensive-language datasets using TF-IDF and logistic regression classifiers. Rendering correctness is measured by an OCR round trip over 200 feature words per language. Default rendering yields character error rates of 0.820 to 0.979, meaning the label no longer carries its token; the common reshape-and-reorder workaround fails for Urdu at 0.998, worse than no correction; and the Matplotlib 3.11.0 text rewrite inverts that workaround, while SHAP-RTL remains correct under both versions. The framework also verbalizes the same attributions as short contextual explanations in the reader's language, constrained to the identified features. Evaluation in this paper concerns rendering correctness; assessment of the generated explanations is left to future work. The study highlights the importance of language-aware visualization in making post hoc explainability more accessible across different writing systems. --- ### Don't Read the Log: Execution Traces Contaminate Verifiers in Video-Generation Agents Authors: Jian Xu Published: 2026-09-23 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.28564v1 Agentic video-generation systems close a loop between a generator and a verifier: an LLM plans shots, calls a text-to-video model, and a multimodal judge decides whether the result satisfies the request. To diagnose where a long workflow fails, recent harnesses deliberately show the judge more than the video-the agent's execution trace, its plan, the narration it synthesized. We ask whether this auxiliary text moves the judge's verdict on purely \emph{visual} requirements, holding the frames fixed. On a benchmark of 109 generated two-event clips with manual labels, in which the requested event is either visibly completed or visibly missing, a trace that reports a successful tool call makes three open-weight Qwen-VL judges (7B, 8B, 32B) accept $78$--$90\%$ of the failures, up from $7$--$19\%$ without text, and a contradicting trace makes them reject up to $100\%$ of correct clips; an instruction to ``use only the frames'' does not remove the effect. Frontier closed judges are essentially unmoved on the same clips, showing that the vulnerability is a property of the judge's learned trust in tool logs rather than of the task. Plan-derived text carries no clip-specific information, so it can only shift a judge's operating point, and in a repair loop that shift becomes a cap on the true pass rate that no repair policy can exceed; the cap matches simulation to two decimals. In the loop, contamination is exploited without any adversarial agent: an honest LLM planner that always regenerates ends with a judge pass rate of $1.00$ and a human-labelled pass rate of $0.28$, and a pipeline in which a cheap checker writes its verdict into the trace launders that checker's errors into a stronger final judge ($0.69$ false accepts). --- ### SpaFactor: Lightweight Spatial Context-Aware Gene Program Modeling for Histology-to-Transcriptomics Inference Authors: Shiting Ruan, Xitong Ling, Qiming He, Ziyou Yan, Huaitian Yuan, Tian Guan, Ying Xiao, Xu Guan, Yonghong He Published: 2026-09-23 Categories: cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2609.28563v1 Spatial transcriptomics (ST) profiles gene expression within tissue architecture, but its cost and experimental complexity limit routine use. Predicting spatial expression from routinely available hematoxylin and eosin (HE) images therefore offers a scalable alternative. However, conventional methods often fit high-dimensional gene outputs as independent targets, overlooking the biological coordination among genes while remaining vulnerable to high-dimensional noise and overfitting. Existing attempts to address this limitation often rely on computationally heavy graph networks or complex auxiliary supervision. We therefore introduce SpaFactor, a lightweight and efficient low-rank morphology-program-gene factorization framework. At the input, SpaFactor efficiently fuses the visual representation of the central spot with multiscale local and regional neighborhood context, yielding a histologic representation that captures cellular morphology and microenvironmental heterogeneity. For modeling, a residual MLP stably learns a nonlinear mapping from the tissue microenvironment to low-dimensional latent gene programs. These activities are decoded through shared gene loadings into coordinated multi-gene expression predictions. Across five public cohorts, SpaFactor achieves the best aggregate performance, with particularly clear improvements for spatially variable genes, and more faithfully recovers biologically organized spatial patterns. These results demonstrate that lightweight joint modeling of tissue context and gene programs can improve both predictive accuracy and biological fidelity. --- ### Matrix Aggregation Operators Authors: Inmaculada Gutiérrez, Asier Urio-Larrea, J. Tinguaro Rodríguez, Daniel Gómez, Javier Montero, Humberto Bustince Published: 2026-09-23 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.28562v1 Aggregation theory has traditionally focused on operators defined over vectors. However, many applications-including Multi-Criteria Decision Making, Group Decision Making, Fuzzy Rule-Based Classification Systems, and overlap/grouping indices-require aggregating information naturally structured as a matrix of membership degrees (e.g., where a set of objects interacts with a family of fuzzy sets). Despite this, no formal framework has been proposed for this class of operators, partly due to the common practice of flattening matrices into vectors (which discards structural information) and partly due to a reliance on decomposable operators that aggregate rows and columns sequentially. This paper addresses this gap by formalizing the notion of a matrix aggregation operator (MAO). We analyze the decomposability and symmetry properties of MAOs, showing that certain operators cannot be expressed in decomposable form and examining several notions of symmetry. Finally, we introduce a family of MAOs termed maximum entropy global coverage indices (MEGCIs), provide a construction method for them based on combining grouping functions and MEOWA operators, and illustrate their usefulness in cluster quality assessment through an extensive computational study. --- ### CARE: Condition-Aware Representation Regularization for Diffusion Models Authors: Fengjia Guo, Zhuoyi Yang, Jie Tang Published: 2026-09-23 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2609.28561v1 Recent advances in diffusion models highlight the importance of representation regularization for improving sample quality and training efficiency. However, commonly used regularization methods often overlook the built-in conditions (such as labels or texts) which directly determine the generation target. In this work, we demonstrate how conditioning signals affect the feature distribution and introduce the CARE (Condition-Aware REpresentation regularization). CARE is a lightweight plug-and-play regularization framework that dynamically modulates feature distribution based on condition similarity. CARE leverages built-in conditioning signals to judiciously guide the representation space, promoting tighter feature clusters for similar conditions without relying on explicit alignment losses or external supervision. Empirically, CARE consistently improves both visual fidelity and convergence stability across both class-to-image and text-to-image tasks. On ImageNet, CARE achieves a 19.08\% reduction in FID in 400k training steps, leading to a 3.5$\times$ speed-up. When applied to text-to-image generation, CARE lowers FID by 16.61\% in 200k iterations and improves semantic alignment between generated samples and text prompts. Moreover, CARE can be seamlessly integrated with existing regularization methods, yielding additional performance gains. --- ### Speculative Evaluation of Stochastic LLMs Authors: Qianli Shen, Xiang Li, Ruomeng Ding, Yanxi Chen, Daoyuan Chen, Yaliang Li Published: 2026-09-23 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.28560v1 Evaluating a stochastic large language model is costly: benchmark scores estimate expected performance from randomized rollouts, yet uniform repetition ignores sharp differences in task-level rollout variance. We ask how to minimize the variance of a fixed-benchmark mean under an exact rollout budget. We develop Speculative Evaluation with a Hierarchical Bayesian Neyman (HBN) policy with pilot size and stage weight jointly chosen ex ante. It runs a short uniform pilot, pools per-task success counts with a hierarchical Bayesian model, and uses posterior expectations of task-level sampling variances for exact positive-integer Neyman allocation. To mitigate the pilot synchronization barrier, HBN-async speculatively executes continuations from partial pilot feedback and retains those selected by the final allocation. Across six checkpoints and 18 benchmark groups, we evaluate 107 nondegenerate benchmark-checkpoint profiles. For rollout budgets of 8-64 per task, Speculative Evaluation reduces variance relative to Uniform by 12.8%-33.6% on average across profiles, outperforming hindsight-tuned empirical and independent Bayesian baselines. Real-generation experiments that account for the pilot synchronization barrier show that HBN-async mitigates its overhead, helping translate statistical efficiency into practical evaluation benefits. --- ### Who Is Behind the Harness? Fingerprinting LLMs through Agentic Behavior Authors: Chuyi Wang, Xiaohui Xie, Tongze Wang, Fangchen Luo, Yong Cui Published: 2026-09-23 Categories: cs.CR, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2609.28559v1 LLMs increasingly operate through coding-agent harnesses that inspect repositories, invoke tools, and modify files. Substituting the model behind such an agent can therefore change security-relevant decisions, including whether it verifies changes or recovers safely from failures. Existing LLM fingerprints largely infer identity from direct text or token distributions. In coding agents, these signals are mediated by system instructions, controller logic, tools, and execution feedback, limiting their transfer. We present LIDAR (LLM Identification from Decisions and Actions at Runtime), an active black-box fingerprinting method for coding-agent execution. Three coding probe pairs expose post-edit verification, transient-failure recovery, and specification--test conflict resolution under controlled changes. LIDAR represents the resulting trajectories with complementary instance-level and distribution-level features and compares them with clean references using a lightweight probabilistic identifier. It requires no access to model weights, logits, or provider internals. Across 36 models from seven families and two agent harnesses, LIDAR achieves high Top-1 accuracy and MRR and outperforms four existing fingerprinting and API-auditing baselines. Ablations confirm that the two feature levels, all probe pairs, and their controlled variants contribute. These results show that agent execution behavior provides model-identity evidence beyond final outputs. --- ### CFD Correction of Open Tip Clearance Flow in a Compressor Cascade Using VAE Latent Space Adaptation Authors: Xiang Zuo, Hefang Deng, Caiyan Chen, Honglin He, Mingmin Zhu, Songan Zhang, Jinfang Teng Published: 2026-09-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.28558v1 CFD predictions of open tip clearance flow in compressor cascades are subject to discrepancies relative to experiments, while experimental observations are sparse and high-resolution experimental ground truth is unavailable. This study proposes a non-intrusive correction method based on a variational autoencoder (VAE) and latent-space adaptation. A VAE is first trained using a dataset of 166 parametrically sampled CFD total pressure loss fields to learn a low-dimensional statistical representation of these fields. The VAE is then frozen, and a low-rank latent-space adapter is trained using only 12 paired CFD--experiment operating conditions. An observation operator maps the corrected high-resolution fields to the experimental observation space, allowing supervision to be applied only at the available measurement locations and within the measured pitchwise windows. In the current 12-fold cross-validation, the mean absolute error decreases from 0.1335 to 0.0473, the root mean square error from 0.1717 to 0.0621, and the relative $L_2$ error from 0.5108 to 0.1871. These results indicate that the method improves agreement between CFD predictions and sparse experimental observations of open tip clearance flow without modifying the RANS solver or constructing artificial high-resolution experimental labels. --- ### BaseCamp --- An Agentic AI Framework for Automating DNA Sequencing Data Pipelines Authors: Eranga Bandara, Xueping Liang, Asanga Gunaratna, Tharaka Hewa, Abdul Rahman, Peter Foytik, Safdar H. Bouk, Sachini Rajapakse, Isurunima Kularathna, Pramoda Karunarathna, Chalani Rajapakse, Ng Wee Keong et al. Published: 2026-09-23 Categories: cs.AI, q-bio.GN arXiv: https://arxiv.org/abs/2609.28557v1 DNA sequencing pipelines, spanning quality control, alignment, variant calling, and annotation, are now reliably executed by workflow management systems that orchestrate established bioinformatics tools at scale. What remains manual is the decision layer surrounding that execution: selecting quality thresholds appropriate to a sample and platform, adjudicating borderline variant calls, diagnosing anomalies, and determining which findings warrant expert review. These decisions are repetitive, judgment-intensive, inconsistent across operators, and frequently undocumented. This paper introduces BaseCamp, a novel agentic AI framework for automating the decision layer of DNA sequencing pipelines. The framework decomposes the pipeline into six specialized AI agents, covering sample intake and quality control, alignment, variant calling, annotation, cross-stage monitoring, and reporting. Critically, BaseCamp agents do not perform sequence analysis: established tools execute alignment, calling, and annotation, while the agents select among them, configure them, interpret their output, and decide what follows. This confines language model reasoning to the judgment layer where it is reliable and preserves the reproducibility existing tooling guarantees. Agent reasoning is powered by a consortium of fine-tuned, domain-specialized large language models coordinated by a central reasoning LLM, executing locally so no sequencing data leaves the operating environment, under human-in-the-loop orchestration. Evaluation shows agent-generated configurations are concordant with expert practice, that an explicit filtering ledger renders inspectable what filtering otherwise removes without trace, and that cross-stage anomaly detection surfaces conditions execution monitoring misses. BaseCamp offers a generalizable blueprint for agentic automation of scientific data pipelines. --- ### Pistis Technical Report Authors: Heyun Chen, Xiaohan Lan, Jiaxi Li, Zhilin Lu, Qi She, Weiwen Xu, Fei Yu, Yujie Zhong, Jinghuan Chen, Zijian Feng, Siyu Jiao, Yiheng Lin et al. Published: 2026-09-23 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2609.28554v1 We introduce the Pistis model family, comprising 27B- and 9B-parameter multimodal large language models built on Qwen3.6 and Qwen3.5, respectively, and developed through a general and scalable post-training framework. The framework first establishes a strong foundation through large-scale multimodal supervised fine-tuning (SFT). Building on this SFT foundation, we propose Interleaved Distillation and Reinforcement Learning (IDRL), a novel post-training paradigm that tightly integrates on-policy distillation and reinforcement learning within a single training loop. By alternating between the two objectives, rather than optimizing either in isolation or combining them in a static joint loss, IDRL enables more effective knowledge transfer, greater optimization stability, and more precise credit assignment for long-horizon agentic trajectories, leading to stronger performance while mitigating common capability trade-offs. At both model scales, the framework produces two specialized variants: Pistis-Thinking, designed to strengthen deep multimodal reasoning, and Pistis-Agentic, which additionally incorporates agentic trajectory data to support long-horizon planning, iterative reasoning, and tool use. Pistis-Agentic is particularly strong in multimodal search. Both scales outperform their corresponding base models. Beyond model-parameter optimization, we further introduce Pistis-Auto-Harnessing (PAH), a system-level method that automatically improves the agent's inference harness through iterative optimization. Experiments demonstrate that PAH enhances the model performance without updating the model parameters or increasing the interaction budget. --- ### SMILESGNN: Interpretable Clinical Toxicity Prediction via SMILES-Graph Cross-Attention Fusion Authors: Quang Minh Nguyen, Thuy Quynh Nguyen, Duc Minh Le, Ho Nhat Minh Nguyen, Thanh Long Dai Doan, Trong Nghia Nguyen Published: 2026-09-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.28553v1 Drug toxicity prediction is critical for reducing late-stage attrition in drug discovery, yet remains challenging due to severe class imbalance, scaffold-based generalization, and the clinical need for interpretable predictions. Single-modality approaches-SMILES Transformers or graph neural networks capture complementary aspects of molecular structure, while sequence-only models cannot directly provide graph-attributed explanations. We present SMILESGNN, a multimodal architecture that fuses a SMILES Transformer encoder and a GATv2 graph encoder via cross-attention, and SMILESGNN-PT, a variant using a ChemBERTa-2 pretrained backbone. The design retains an explicit graph branch within the predictive pipeline, supporting GNNExplainer-based analysis of substructures associated with toxic predictions. On ClinTox, SMILESGNN achieves AUC-ROC 0.987 and F1 0.906 with only 0.4M parameters, performing competitively with a strong SMILESTransformer and a larger ChemBERTa-2/GATv2 concat-fusion baseline. On Tox21 (12 tasks), SMILESGNN-PT obtains mean AUC-ROC 0.750, comparable to ChemBERTa-2 alone and the same-backbone concat-fusion baseline. Overall, the results suggest that cross-attention is a practical fusion alternative that preserves competitive predictive performance while enabling graph-based interpretability support. --- ### An Order-Theoretic Characterization of Consistent Inductive Inference Authors: Zhou Lu Published: 2026-09-23 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.28551v1 When can a learner make only finitely many prediction errors along every infinite sequence labeled by a fixed, unknown hypothesis? We characterize this form of consistency for arbitrary binary hypothesis classes in ZFC, without requiring a uniform mistake bound. The characterization uses a single linear order on finite realizable traces. Each trace selects its least subtrace, and the order must satisfy two conditions: conflicting traces select different subtraces, and the order is well-founded on the traces of each fixed target. These conditions induce a learner whose selected evidence decreases on every mistake. Conversely, a consistent learner yields such an order through canonical mistake transcripts and the Kleene--Brouwer ordering. The result provides a representation of consistent prediction by finite evidence, answering a question of Lu (2024). --- ### PAWS: Policy-driven Agentic World Simulation Authors: Tiviatis Sim, Jia Hui Woon, Xinming Gao, Chen Gao, Fengbin Zhu, Zheng Huanhuan, Chua Tat Seng, Kenji Kawaguchi Published: 2026-09-23 Categories: cs.AI, cs.CE, cs.MA, cs.SI arXiv: https://arxiv.org/abs/2609.28547v1 Policy interventions propagate through public communication, institutional decisions, and stakeholder responses, yet datasets for financial multi-agent simulation rarely connect these processes to temporally aligned historical evidence. We introduce PAWS, a Policy-driven Agentic World Simulation dataset covering 36 verified U.S. financial and economic policy episodes, 12,727 policy-linked news records, and 65,291 source-grounded stakeholder actions. Each action is linked to its supporting news and represented by a multi-layer event frame capturing its interaction mode, financial-action family and subtype, semantic attributes, and conditional mappings to external taxonomies. Entities are resolved to normalized organizations, and actions are aligned with daily market-return context to support policy-agent simulation replay. On 2,522 stratified action samples, independent AI and human reviewers achieved 89.4% initial agreement on interaction mode, with disagreements subsequently adjudicated. Case studies of the 2008 short-selling ban and 2001 decimalization recover documented policy timelines and associated market patterns across both dense and sparse news settings. A replay study further shows that high accuracy can mask failure to detect rare stakeholder actions, identifying action timing and calibration as central challenges. PAWS provides an auditable substrate for evaluating agent influence, policy-response cascades, and action-outcome alignment in historically grounded financial simulations. --- ### On the Diffusibility of High-Dimensional Latents Authors: Chao Feng, Zhiyang Xu, Bowei Chen, Yuanjun Xiong, Xiyao Wang, Jui-Hsien Wang, Richard Zhang, Zhe Lin, Andrew Owens, Yijun Li Published: 2026-09-23 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.28473v1 Representation Autoencoders (RAEs) enable diffusion models to operate in the feature spaces of pretrained visual encoders. However, many off-the-shelf encoders are not optimized for faithful reconstruction, discarding fine-grained visual details. As expected, finetuning these encoders for image reconstruction recovers such details. However, perhaps counterintuitively, this procedure reduces the effective dimensionality of the resulting representation, and the altered geometry has downstream effects on generation. Specifically, we show that using the standard velocity prediction in flow matching in this high-dimensional space requires the model to fit orthogonal noise directions outside the low-dimensional signal manifold, making optimization inefficient. This motivates using the clean data parameterization ($\boldsymbol{x}_{0}$-prediction) instead, which focuses learning on the underlying signal manifold. Across experiments with multiple strong-reconstruction encoders, we show that $\boldsymbol{x}_{0}$-prediction consistently improves text-to-image generation performance. --- ### Contrastive Learning for Authorship Verification Authors: Peter Kirby Published: 2026-09-23 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.28471v1 Our results show that contrastive learning outperforms a classification-based approach to authorship verification under the tested settings. We identify loss function, batch size, training duration, pre-trained model, input context length, and random text span data augmentation as important factors of model performance. Based on these considerations, we develop a ModernBERT Bi-Encoder model that achieves 98.4% accuracy on the PAN21 authorship verification task. --- ### StudentBench: AI and human tutoring yield equivalent GRE learning gains Authors: Curtis Northcutt, Inaara Hasmani, Kevin Feng, Trevor Khangi, Andreas Plesner, Jonas Mueller Published: 2026-09-23 Categories: cs.AI, cs.CY arXiv: https://arxiv.org/abs/2609.28470v1 Artificial intelligence offers an unprecedented opportunity to augment human capabilities, yet progress at the frontier has focused primarily on advancing model capabilities. We introduce StudentBench, a suite of AI teaching evaluations and a public platform that enables large-scale data collection with over 175,000 student-AI messages to study whether large language models (LLMs) produce learning gains equivalent to human tutoring. Using StudentBench, we measured learning gains on Quantitative and Verbal GRE questions across 2,383 human participants receiving AI tutoring, human tutoring, or no tutoring. We establish that AI tutoring is statistically equivalent to expert human tutoring for GRE learning gains (p = .015), and in five of the seven GRE domains, the best performing AI tutor surpassed the human tutor, on average. In a second study, expert human tutors compared LLM-generated lesson plans and practice problems through 2,028 pairwise rubric evaluations. Together, the two studies clearly separate AI tutors across: (1) lesson planning, (2) practice-problem creation, (3) conversational pedagogy, (4) cost, and (5) engagement. Surprisingly, one AI tutor achieved learning gains equivalent to human tutoring (p = .044) at 918 times lower cost (USD 0.0052 for AI versus USD 4.81 for human, per percentage point gained). For Quantitative GRE sessions, faster AI replies correlated with more student messages, more messages with more correct practice, and more correct practice with larger learning gains (all p < .002). The StudentBench platform is freely available at https://studentbench.org. --- ### Where Should I Join? Robot Group Joining via Language-Guided Goal Prediction Authors: Zilin Fang, Zishuo Wang, Gim Hee Lee, David Hsu Published: 2026-09-23 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.28467v1 Social navigation typically assumes a specified goal and focuses on reaching it while respecting social conventions, whereas robot group joining requires predicting where to join based on the group's real-time activity and formation. This is a highly semantic task, yet an important capability for applications such as robotic guide dogs and autonomous mobility scooters. We formulate language-grounded robot group joining: given an observation and a natural-language description of a target group, the robot identifies the relevant group members and predicts socially compliant joining poses. For grounding, we generate structured candidate subsets through recursive spectral partitioning and rank them with a language-conditioned image--geometry model. Given the grounded group, a goal predictor leverages human-formation priors to produce a multimodal energy--orientation map over feasible robot poses. Experiments on conversations, queues, and audiences across varying group sizes, crowd densities, and visual ambiguities show that our method achieves competitive grounding accuracy with sub-second inference and outperforms all baselines in joining-pose prediction. Real-robot experiments further demonstrate group joining in both static and dynamically changing interactions. --- ### The Past Frames the Future: Memory for Autoregressive Video Generation Authors: Harold Haodong Chen, Rongjin Guo, Disen Lan, Wen-Jie Shu, Hongfei Zhang, Hanzhe Hu, Shengtao Yao, Zixin Zhang, Guibin Zhang, Zhefan Rao, Jinxiu Liu, Yexin Liu et al. Published: 2026-09-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.28466v1 Advances in generative models have improved video fidelity, enabling long-horizon generation, interactive world modeling, and evolving visual environments. Autoregressive (AR) video generation extends visual sequences through causal rollouts. However, a fundamental bottleneck emerges: as the generated sequence expands, practical models must operate under strictly bounded context windows, storage, and computational limits. Consequently, critical historical information, e.g., entity identities, dynamic states, and intervention-induced causal changes, often leaves the active context long before its relevance diminishes. Overcoming this limitation and maintaining temporal persistence constitutes a fundamental memory problem. We present a systematic and comprehensive review of memory mechanisms in AR video generation. We formulate memory operationally as persistent historical information maintained across outer AR steps, capable of influencing future generation even after the originating evidence is no longer locally accessible. Building upon this unified framework, we organize the literature through five complementary perspectives: (I) Forms, the representational carriers of history; (II) Functions, the specific semantic and physical information requiring preservation; (III) Operations, the lifecycle of writing, reading, updating, managing, and integrating memory; (IV) Learning, the optimization of memory behaviors under closed-loop rollouts; and (V) Evaluation, the paradigms for diagnosing genuine memory capabilities. We conclude by synthesizing open challenges, including composable and resource-aware memory architectures, trustworthy state updating, self-rollout learning, and standardized evaluation. By bridging representations, mechanisms, and learning paradigms, this paper establishes a structured foundation for developing reliable, memory-conditioned video generation systems. --- ### Even Sharper Bounds for Transductive Learning and Its Applications Authors: Yingzhen Yang Published: 2026-09-23 Categories: cs.LG, cs.IT, math.ST arXiv: https://arxiv.org/abs/2609.28459v2 We introduce Sharper Transductive Local Complexity (STLC), a localized complexity method for transductive learning under uniform sampling without replacement. The construction starts from a Bernstein-type concentration inequality for the supremum of the test--train empirical process. Its proof uses the modified log-Sobolev inequality for the swap walk and a two-parameter entropy closure. A peeling argument with a surrogate localization functional then gives excess-risk bounds with the same fixed-point and confidence terms as the classical inductive local Rademacher-complexity bounds, without the additional logarithmic confidence factor in earlier transductive results. For realizable learning over a binary class of VC dimension $\dVC$, with training size $m$, test size $u$, and $u\ge m\ge\dVC$, STLC yields $\cO\{\dVC\log(me/\dVC)/m\}$. This matches the standard inductive rate and, when $m\ge9$, is within a logarithmic factor of the transductive minimax lower bound of order $\dVC/m$. For transductive kernel learning, STLC gives a spectrum-adaptive excess-risk bound without the multiplicative imbalance factors appearing in the earlier local-complexity bound. ---