# Research Papers — 2026-09 · bundle page 256 of 266 > Full-text markdown of papers #12751–12800 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/256 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/255.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/257.md) ### Don't Just Look, Intervene: Perturbation Based Region Labeling for VQA Images Authors: Marko Jojic, Zhaonan Li, Ben Zhou Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.13228v1 Vision Language Models (VLMs) should rely on visual evidence that directly determines the correct answer, but supervision for grounding visual reasoning is often expensive to obtain manually or tied to dataset-specific annotation primitives. We instead introduce model-causal visual evidence as an annotation target, defined as the set of image regions whose counterfactual intervention changes a model's answer distribution for a given image-question pair. Based on this principle, we introduce Counterfactual Search for Grounding Regions (CSGR). CSGR is a scalable pipeline that proposes candidate regions, perturbs them, measures their effect on answer sensitivity, and aggregates this evidence across multiple judges to approximate answer-critical regions in VQA data. To assess whether CSGR annotations contain a useful supervision signal, we plug them into three existing grounding-aware training routines: attention steering, Visual CoTfinetuning, and latent visual reasoning. These experiments test whether the proposed annotation scheme can provide a useful supervision signal across multiple ways of consuming region labels, rather than introducing a new way of using them. Across competing automatic region-labeling mechanisms, CSGR annotations provide the most consistent gains over Cross Entropy-only finetuning in both in-domain and out-of-domain evaluations, indicating that the proposed labeling scheme captures useful region-level information. --- ### Synthetic Leprosy Image Generation Using Mask-Conditioned Latent Diffusion and Transfer Learning from Large Chronic Wound Datasets Authors: Yusuf Abdulkadir Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.13226v1 Machine learning for neglected tropical diseases is limited by data, not algorithms: public annotated image sets for leprosy (Hansen's disease) number in the hundreds, orders of magnitude below what generative models require. We ask whether a model trained on abundant chronic wound photography transfers to this low-data regime. We build a three-stage pipeline. First, a DeepLabV3-ResNet50 segmentation network (validation Dice 0.876, IoU 0.799) supplies lesion masks for two wound datasets that ship without them. Second, we assemble a mask-conditioned latent diffusion model from Stable Diffusion 1.5 components and train it on 3,280 region-of-interest wound crops, widening the UNet input convolution from 4 to 11 channels to admit three mask feature maps and a blurred low-frequency context latent. Third, we fine-tune this model on 708 leprosy image-mask pairs drawn from 764 images of approximately 150 patients. We evaluate with LPIPS perceptual distance, anchored by a real-versus-real baseline computed on the same 242 anchor images as the cross-set comparisons; without that reference the cross-set distances cannot be interpreted. The generated set shows no mode collapse: its internal perceptual diversity (0.662) is statistically indistinguishable from that of the real leprosy set (0.672, 95% CI (0.664, 0.680)). Generated images sit 0.044 LPIPS outside the real distribution - measurably apart, but under half of one standard deviation. Fine-tuning shifted the output distribution only marginally, which we trace to lesion geometry reaching the network through input concatenation alone. Chronic wound photography is therefore a viable donor domain for leprosy lesion synthesis: low-level appearance transfers well, and the remaining barrier is semantic control rather than image quality. --- ### Part Grounding, Not Action Knowledge: Locating the Bottleneck in VLM Affordance Prediction Authors: Sarthak Sattigeri Published: 2026-09-01 Categories: cs.CV, cs.LG, cs.RO arXiv: https://arxiv.org/abs/2609.13225v1 Benchmarks agree that vision-language models reason poorly about low-level manipulation, but an aggregate accuracy score does not say which step fails. We separate two steps that affordance questions conflate: identifying which part of an object to act on, and knowing what action that part requires. Across 19 articulated objects we asked eight models, spanning three developers, what motion a robot should apply. Under an open prompt, push was produced once in 64 evaluations where it was correct, despite being correct for 8 of 19 objects and appearing in the offered label set every time. Inspecting the outputs showed why: models described a different part than the one being scored, e.g. explaining how to pick up a camera rather than press its button. Naming the target part raises action accuracy by 0.32 to 0.63 for every model, from 0.158-0.474 to 0.684-0.947, and push recall from 0-1/8 to 7-8/8. No model beats a constant answer that ignores the image under the open prompt; once the part is named, all eight do. Asked to describe the same part in free prose with no label set, models produce pressing language for 6 to 8 of 8. These results are hard to reconcile with missing action knowledge, and instead point to part grounding as the dominant bottleneck, a pattern that holds across all three model families and does not diminish with capability. Naming the part supplies the grounding variable, so this bounds what a perfect part detector would offer rather than demonstrating a general model of mechanics. Two supporting results agree: on real photographs only three of eight models localize grasp points better than a constant baseline, and on rendered objects none do. We also document two measurement errors of our own, a threshold that let a constant baseline score 0.929 and a labelling rule wrong on 4 of 19 objects, both caught only by testing our numbers against trivial alternatives. --- ### Reliability-Aware Hybrid-K Ensemble Selection for Cervical Cytology Classification: Integrating Discrimination, Calibration, and Selective Prediction Authors: Nisreen Albzour, Sarah S. Lam Published: 2026-09-01 Categories: eess.IV, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2609.09189v1 High classification accuracy alone is insufficient for clinical image analysis, where calibrated confidence and reliable uncertainty estimates are essential. This study proposes a reliability-aware Hybrid-K ensemble selection framework for multiclass cervical cytology classification using the SIPaKMeD dataset. Nine deep learning architectures were evaluated using a fixed stratified five-fold partition and three training seeds. After post-hoc temperature scaling, models were assessed using macro-F1, accuracy, AUROC, expected calibration error (ECE), worst-class ECE (WC-ECE), area under the risk-coverage curve (AURC), Brier score, and negative log-likelihood (NLL). Models were ranked using an equal-weight composite score, and Hybrid-K ensembles were formed from the top-ranked models using soft voting. Robustness was examined using 5,000 Dirichlet-sampled metric-weight vectors, leave-one-metric-out analysis, and corrected paired testing across 15 fold-by-seed evaluations. The final Hybrid-2 ensemble, comprising Swin-Tiny and TinyViT-5M, reduced AURC by 43%, NLL by 17%, and WC-ECE by 36% relative to the best individual model. It was selected in 96.8% of random weighting scenarios, remained unchanged across all leave-one-metric-out analyses, and improved the full composite score. However, per-metric gains were not statistically significant after Holm-Bonferroni correction (all adjusted p >= 0.168). Because post-hoc calibration did not use a fully independent calibration set, calibration-dependent results should be interpreted as exploratory internal estimates. Overall, the framework identified a compact ensemble robust to alternative metric weightings and improved reliability point estimates under internal validation on a single dataset. --- ### DART: A DAG-Based Reputation and Incentive Framework via Blockchain-Enabled Governance for Trustworthy LLM Multi-Agent Collaboration Authors: Manoj Kumala, Xinyun Liua, Ronghua Xu Published: 2026-09-01 Categories: cs.MA, cs.AI arXiv: https://arxiv.org/abs/2609.05529v1 Large language model (LLM)-based multi-agent systems (MAS) predominantly rely on centralized orchestration and lack formal verification mechanisms for agent reliability, participation, and system-level behavioral alignment. These shortcomings leave open environments severely vulnerable to uncooperative or malicious agents. This work proposes DART, a Directed Acyclic Graph (DAG)-based reputation and incentive regulation framework for trustworthy multi-agent collaboration, combining centralized operational orchestration with blockchain-enabled decentralized governance and accountability. DART unifies DAG workflow orchestration, capability and reputation-aware task allocation, dynamic behavior updates, multi-factor incentives, and smart contract accountability paired with IPFS storage. Under this paradigm, agent selection dynamically balances task alignment, historical reputation, and workload, while post-execution behavioral evidence continuously calibrates agent trust and the probability of future participation. Evaluated across four axes, DART achieves 93.6% Pass@1 on GSM8K and builds a full-stack application in 142 s using two agents, outperforming centralized baselines. Across five independent 150-round longitudinal trials, Full DART achieves a mean task success rate of 93.33 +/- 2.26%, output quality of 0.9357 +/- 0.0117, retry rate of 0.2307 +/- 0.0816, and allocation delay of 1.1153 +/- 0.0408 s, consistently outperforming its ablated configurations DART isolates persistent and intermittent malicious agents, obtaining a 99.3% output containment rate and restoring system success to 99.8%. These results demonstrate the potential of coupling reputation, incentives, DAG-based coordination, and verifiable blockchain-enabled governance to support adaptive and accountable multi-agent collaboration. --- ### Beyond "AI Helps Humans": Decision-Targeted Evaluation Design for Human-Agent Teams in the Agentic Era Authors: Hamed Khosravi, Xiaoming Huo Published: 2026-09-01 Categories: cs.AI, cs.HC arXiv: https://arxiv.org/abs/2609.05527v1 Wherever a coding agent works under engineer supervision, or a clinical model assists a radiologist, the deployment question is whether to keep the human-AI workflow or replace it with the human alone or the agent alone. The human-AI workflow is worth keeping only if it beats both of those alternatives. Yet once it is deployed, neither alternative outcome is observed: recovering one means replaying the task under that alternative, and every replay costs expert time or compute. Under a fixed replay budget, the design question is therefore which tasks should be more likely to receive a human-only replay, and which an agent-only replay. Existing methods do not directly target this decision. Agent benchmarks do not choose which missing baseline to measure, variance-based sampling ignores which of the two comparisons is closer to failing, and Bayesian information methods focus on learning model parameters instead of making the deployment decision. We propose TEAM-Design, a rule that gives every task two replay probabilities, one per baseline. It raises a probability where the missing baseline outcome is hard to predict from what is already known about the task and where that comparison is harder to establish, and lowers it where replay is expensive. We prove that the rule solves this budgeted design problem, and that drawing the replays at random from recorded probabilities still controls the chance of wrongly declaring that the workflow beats both. We reanalyze 6 clinical settings, where no human-AI workflow beats both alternatives, and a coding benchmark, where one does, then evaluate TEAM-Design on synthetic designs and on a semi-synthetic design built from a real chest X-ray reader study. TEAM-Design works best when one of the two comparisons is clearly harder to settle than the other, and can do worse than variance-based allocation when the two are similarly difficult. --- ### DIVA: Exploiting Cross-Step Conditional Propagation for Visual Jailbreaks in Discrete Diffusion Vision-Language Models Authors: Guorui Song, Runqing Tang, Jingye Zhang, Luyuan Zhang, Feice Huang, Cong Ray, Guocun Wang, Dake Zhong, Choo Sin Wai, Bingquan Dai, Chuming Wang, Tongxu Lin et al. Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.05525v1 Large vision-language models (VLMs) are increasingly deployed in safety-critical settings, yet existing visual jailbreak research has focused almost exclusively on autoregressive architectures, leaving an important emerging family unstudied: multimodal discrete diffusion vision-language models (dVLMs). We identify a vulnerability specific to diffusion generation: because the visual embedding conditions every reverse denoising step rather than acting as a one-time prefix, adversarial visual semantics are repeatedly propagated and amplified across the generation trajectory, a phenomenon we term cross-step conditional propagation. We provide empirical evidence through stage-sensitivity analysis, prompt-level switch rates, and pairwise denoising-bin disagreement metrics, confirmed by bootstrap resampling. We propose DIVA (Discrete-diffusion Vision-language model Attack), a white-box visual jailbreak framework using cross-modal intent obfuscation and diffusion-aware multi-timestep adversarial optimization. Across three dVLMs, DIVA reaches 58.8%, 67.7%, and 69.1% HADES ASR under the Beaver reward-model metric, outperforming visual jailbreak baselines designed for autoregressive models. Code: https://github.com/loststars2002/DIVA --- ### Infrastructure-based Monocular 3D Vehicle Localization Framework with Experimental Validation Authors: Akos T. Kopeczi-Bocz, Tian Mi, Gabor Orosz, Denes Takacs Published: 2026-09-01 Categories: cs.CV, cs.LG, eess.IV arXiv: https://arxiv.org/abs/2609.05523v1 This paper presents a one-stage learning framework that maps monocular roadside-camera images directly to vehicle states in a ground-fixed coordinate frame. Unlike conventional approaches that first detect vehicles in the image plane and subsequently apply geometric post-processing, the proposed method leverages features from a pretrained object detector to jointly estimate each vehicle's ground-plane position, dimensions, and yaw angle. The framework therefore uses visual features not only for vehicle detection but also for direct spatial and orientation estimation. To support model training and evaluation, we develop a data-collection and label-generation pipeline based on synchronized video from a roadside camera and an unmanned aerial vehicle (UAV). Acting as a temporary top-view sensing platform, the UAV provides vehicle trajectories, dimensions, and orientations, which are transformed into the ground-fixed coordinate frame and temporally aligned with the roadside-camera images to generate ground-truth labels. The framework is evaluated using data collected during multiple experiments at the Mcity Test Facility. Results show that the proposed method can recover vehicle trajectories and orientations from monocular roadside imagery without a separate geometric post-processing stage, demonstrating its potential as a scalable approach to infrastructure-based perception at urban intersections. --- ### Diffusion models for eye-gaze trajectory generation using position and velocity representations Authors: Laxman Basnet, Alexander Szorkovszky, Pedro G. Lind, Anis Yazidi, Shailendra Bhandari Published: 2026-09-01 Categories: cs.CV, cs.AI, cs.NE arXiv: https://arxiv.org/abs/2609.05522v1 Eye-tracking data are expensive to collect, requiring specialized hardware and controlled laboratory conditions, and difficult to share because of privacy constraints. We address this using two complementary denoising diffusion probabilistic models (DDPMs) for unconditional generation of eye-gaze dynamics from visual-search data. Both use an identical FiLM-conditioned one-dimensional U-Net with self-attention (19.35,M parameters), trained on 8,s sliding-window sequences from 28 participants. One model generates raw two-dimensional gaze-position sequences, while the other generates two-component velocity sequences; each uses representation-specific preprocessing, training settings, data partitions, and evaluation protocols. Both are evaluated across three independent training seeds, with aggregated metrics reported as mean,$\pm$,SD. The position-space model achieves a mean Jensen-Shannon (JS) divergence of $0.016\pm0.004$ across nine kinematic features, with the highest feature-wise mean below $0.030$, fixation duration within 2% of real data, and a Fr'echet Gaze Distance more than an order of magnitude below statistical and Markovian baselines. Under a Train-on-Synthetic-Test-on-Real protocol, synthetic-only training achieves $R^2=0.66\pm0.02$, or 82.7% of the real-data $R^2$ point estimate. The velocity-space model achieves a mean JS divergence of $0.0065$ across velocity components, speed, log-speed, and turning angle, with a maximum of $0.015\pm0.005$. Reconstructed path length is less accurate ($0.21\pm0.02$ versus $0.03\pm0.01$ in position space), although the protocols differ. Overall, unconditional diffusion captures local gaze kinematics and short-range temporal and directional structure, while long-range properties such as saccade counts and cumulative path geometry remain targets for future conditioned models. --- ### Situation Awareness for Intelligent Data Distribution in Connected Vehicles Authors: Falk Dettinger, Akshay Narla, Michael Weyrich Published: 2026-09-01 Categories: cs.RO, cs.AI, cs.CV, cs.NI, eess.SY arXiv: https://arxiv.org/abs/2609.05521v1 The limitations of on-board sensors and blind spots caused by occlusion cause the reduction of perception quality in autonomous vehicles. In such cases, cooperative perception provides additional data via Vehicle-to-Everything communication to enhance local perception, causing a large volume of data transmission. The vehicle can focus on acquiring and utilizing relevant data according to the prevailing road context by identifying the current traffic situation. To achieve this, we propose a concept for the situation identification of the vehicle using Bird's-Eye-View images. Firstly, the situation around the vehicle is identified using object detection with semantic segmentation, followed by understanding the context of the traffic using a situation identification module consisting of an open-source projective transformation network Cam2BEV and a situation identification neural network. The concept was evaluated and validated by running the software on the CARLA simulator using the in-built RGB camera and the semantic segmentation camera. Additionally, the portability of the situation identification module for real-world applications was verified on Cityscapes and nuScenes urban driving datasets. Overall, the proposed situation identification approach enables efficient sensor data management by prioritizing relevant data to the current traffic situation. The source code is available in the following link: https://github.com/akshaynarla/DySi_Select --- ### Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images Authors: Jihyun Lee, Hangil Park, Yongmin Seo, Taewon Min, Joodong Yun, Jaewon Kim, Tae-Kyun Kim Published: 2026-09-01 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.05520v1 In this paper, we tackle automatic anomaly detection in multi-illumination and multi-focus display images. The minute defects on the display surface are hard to spot out in RGB images and by a model trained with only normal data. To address this, we propose a novel contrastive learning scheme for knowledge distillation-based anomaly detection. In our framework, Multiresolution Knowledge Distillation (MKD) is adopted as a baseline, which operates by measuring feature similarities between the teacher and student networks. Based on MKD, we propose a novel contrastive learning method, namely Multiresolution Contrastive Distillation (MCD), which does not require positive/negative pairs with an anchor but operates by pulling/pushing the distance between the teacher and student features. Furthermore, we propose the blending module that transforms and aggregate multi-channel information to the three-channel input layer of MCD. Our proposed method significantly outperforms competitive state-of-the-art methods in both AUROC and accuracy metrics on the collected Multi-illumination and Multi-focus display image dataset for Anomaly Detection (MMdAD). --- ### FrOGS: Discrete Neural Sampler for Independent Alloy Configurations Across Chemical Conditions Authors: Kyucheol Min, Elyssa Hofgard, Tess Smidt Published: 2026-09-01 Categories: cond-mat.mtrl-sci, cond-mat.stat-mech, cs.LG arXiv: https://arxiv.org/abs/2609.02948v2 Predicting the thermodynamic properties of an alloy requires sampling its configurations across many chemical conditions and recovering free energies on a common absolute scale. Markov chain Monte Carlo (MCMC) is the standard tool, but it requires separate simulations at different conditions, and auxiliary free-energy methods such as thermodynamic integration are used to place results on a common absolute scale. Modern discrete neural samplers typically use reverse KL divergence as the objective and can be mode-seeking or biased. We present Free energy Offering Generative Sampler (FrOGS), a hybrid discrete neural sampler that couples an autoregressive model to a continuous-time Markov chain (CTMC) to be trained jointly under a single shared loss. FrOGS draws i.i.d. configurations, returns an unbiased estimate of the partition function, and gives consistent estimates of thermodynamic observables. We train a single model across a wide range of chemical conditions to produce estimates on a common absolute free-energy scale. FrOGS matches exact finite-size results on the 2D Ising model and reference phase diagrams for AgPd and CuAu, without mode collapse. We additionally compare to SEGAL, a published autoregressive baseline, and find that only FrOGS recovers the stability range of the CuAu$_3$ phase. --- ### Privacy-Preserving Heterogeneous Multi-LLM Federated Inference for Cognitive Diagnosis Authors: Yagna Manasa Boyapati, Chong Yu, Tianyu Jiang, Justin Zhan Published: 2026-09-01 Categories: cs.CR, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.02947v1 Significant challenges remain in AI-driven educational systems in balancing privacy preservation with accurate cognitive diagnosis. To overcome this, we propose a federated inference framework in which several commercial LLM APIs collaborate without requiring access to raw student data or proprietary model internals. Using multiple federated entities, such as LLaMA-3.3-70B, GPT-4o-mini, and Claude-3-Haiku, our framework builds upon a heterogeneous multi-LLM architecture. The predictions generated by these entities are combined with epsilon-local differential privacy by adding Laplace noise locally to each entity's prediction output before aggregation, while residual-based aggregation mitigates model heterogeneity. Our approach is predicated on an honest-but-curious trust paradigm in which API providers are presumed not to abuse submitted queries, and our differential privacy mechanism shields the published diagnostic results from external inference. We conduct rigorous privacy-utility analysis showing strong privacy guarantees with minimal accuracy loss, and extensive real-world evaluations across three educational benchmarks confirm the framework's practical usability and cross-domain generalizability. --- ### Reflect-SQL: A Self-Reflection Based Framework for Text-to-SQL Authors: Anupreksha Jain, Manish Shrivastava Published: 2026-09-01 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2609.02944v1 Democratizing data access through natural language is a crucial goal for modern enterprises, but the practical adoption of Text-to-SQL is critically hindered by real-world complexities: 1. Obscure and large database schemas, 2. Ineffective retrieval of relevant tables and columns due to structured setting of schemas and vague user query, 3. Generation of syntactically or logically flawed SQL due to a lack of robust validation and correction mechanism. To address these systemic challenges, we introduce Reflect-SQL, a novel framework for Text to SQL, grounded in multi-stage self-reflection approach to develop understanding of obscure schema using a knowledge base, setup a process for effective retrieval and system to generate syntactically/semantically SQL. Instead of a single-pass attempt, our system employs an LLM-as-a-judge driven scoring mechanism within interconnected feedback loops to iteratively refine the results at every stage. A feedback-driven retrieval loop refines the user's natural language query, while a synthesis loop validates and corrects the SQL and finally, an entailment loop optimizes the end-to-end process and continuously enriches the knowledge base. By integrating these layers of reflection, Reflect-SQL bridges the critical gap between user intent and complex data. On the challenging BIRD benchmark, our framework achieves an execution accuracy of 72.03%, significantly outperforming state-of-the-art baselines, demonstrating a major leap in reliability for enterprise applications. --- ### Physics-Driven Independent Pair Generation for Iterative Self-Supervised Low-Dose CT Denoising Authors: Xianlei Han, Shaoyu Wang, Jiancheng Fang, Weiwen Wu, Qiegen Liu Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.02654v1 Low-dose computed tomography (LDCT) measurements contain mixed Poisson-Gaussian noise. However, most self-supervised methods rely on generic image statistics and do not explicitly model this noise, which may limit their ability to effectively suppress realistic LDCT noise. To address this issue, we propose a physics-driven framework with cross-domain iteration for self-supervised LDCT denoising. The proposed framework proceeds in three main steps. First, a learned sinogram prior and the LDCT noise model guide posterior inference of photon counts, enabling separation of the Poisson and Gaussian components. Second, the separated Poisson and Gaussian components are respectively processed by binomial thinning and Gaussian data thinning to construct two branches, and residual scaling matches each branch's noise level to that of the observation, yielding a training pair with approximately independent noise realizations from one low-dose measurement. Finally, the pair is used to train an image-domain network whose forward-projected outputs update the prior. Through cross-domain iteration, the prior and the training pair are progressively refined while maintaining consistency with CT acquisition physics. Experiments on simulated data from AAPM, LIDC-IDRI, and LoDoPaB-CT and on real LDCT data show consistent gains over the evaluated self-supervised baselines across dose levels, with performance comparable to the evaluated supervised baseline. --- ### Convergence Theory of Knowledge Distillation in Asynchronous P2P Gossip Learning Network Authors: Lucas Qingyang Fang, Tiyao Liu, Jinhao Jing, Zeji Li, Kaijie Chen, Harikrishna Kuttivelil, Katia Obraczka Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01952v1 Decentralized, serverless learning increasingly connects devices running different architectures, where the standard tool, decentralized SGD, is undefined as models with different parameter counts cannot be averaged. Knowledge distillation (KD) exchanges soft predictions rather than weights and sidesteps this obstacle, yet convergence theory for fully decentralized, asynchronous peer-to-peer (P2P) KD is lacking. We provide one, relocating consensus from parameter space to function (output) space: a KD event is a geometric contraction operator in logit space on the peers' predictive distributions, which we analyse in the Hilbert space of predictions on a reference measure. Under standard smoothness/variance assumptions and two realizability assumptions, one bridging parameter SGD to the functional step and one controlling restricted task/KD alignment, the time-averaged functional stationarity and function-space disagreement converge at rate $O(1/(ηT))$ to an $O(η)+O(B_f^2)+O(ζ_f^2)$ neighbourhood. Here $B_f$ is the distance from the task optimum to the peers' reachable classes and $ζ_f$ measures persistent local-task heterogeneity. Across homogeneous, width-heterogeneous, and mixed-family networks of the experiments, KD contracts function disagreement by $40-61\times$, while isolated training does not. The sampled stationarity diagnostic has late transient exponents $0.99-1.90$ on the shared-skeleton main runs, and the four-point step-size sweep exhibits the predicted transient: neighbourhood tradeoff. --- ### On-Policy Distillation Meets Off-Policy GRPO: Training Compact Instruction-Following Rerankers Authors: Vignesh Prabhakar, Jialing Pan, Anil Babu Ankisettipalli Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01947v1 Compact instruction-following rerankers are attractive for deployment, but conventional distillation pipelines typically train students by offline imitation of teacher outputs on a fixed set of examples, constraining supervision to the teacher's observed ranking space. We revisit reranker distillation through the lens of reinforcement learning. We propose a two-stage framework combining off-policy teacher optimization with on-policy student distillation. In Stage 1, a 4B teacher reranker is strengthened with off-policy GRPO using LLM-judge feedback on 88K instruction-following examples. In Stage 2, a compact 1B student samples rankings from its own policy and receives soft teacher-derived rewards on those rankings, coupling student exploration with knowledge transfer. Our strongest gains appear under distribution shift. On MAIR-11, the original 11-subset, 869-query evaluation, the proposed student reaches 0.7670 nDCG@6, outperforming offline listwise KD by +4.6 points. Controlled comparisons against offline pairwise RankNet KD and on-policy GKD show that neither changing the offline distillation objective nor moving teacher-distribution matching on-policy reproduces the performance of reward-based on-policy distillation over student-sampled rankings. The advantage persists on MAIR-Full: across all 126 tasks and 9,356 queries, the proposed method obtains the highest task-macro point estimates among the evaluated distillation variants, reaching 0.6808 nDCG@6 and 0.7865 MRR@6. It also exceeds two released 7B RL-trained rerankers on the comparable MAIR-11 evaluation, while the same Stage 2 training procedure consistently improves three architecturally distinct alternative student backbones. On the 9,861-query validation benchmark, the resulting 1B reranker achieves 0.7624 nDCG@6 while providing a favorable quality-efficiency tradeoff relative to larger alternatives. --- ### Pushing Forward Multi-Secret-Key Homomorphic Encryption for Private Average Aggregation Authors: Miguel Morona-Mínguez, Fernando Pérez-González, Alberto Pedrouzo-Ulloa Published: 2026-09-01 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.01945v1 Federated Learning enables multiple clients to train a shared model while keeping their local datasets isolated. However, the exchanged model updates may still leak sensitive information, making private aggregation a central building block in practical deployments, especially in the cross-silo setting. Homomorphic Encryption naturally fits the client--aggregator communication pattern of Federated Learning, but conventional single-key deployments rely on strong non-collusion assumptions. Multiparty Homomorphic Encryption removes this limitation, although recent attacks under restricted decryption access require large-variance smudging noise during collaborative decryption, which significantly increases ciphertext size and implementation complexity. In this work, we propose lightweight multi-secret-key protocols for private average aggregation based on RLWE-based Homomorphic Encryption. Our construction departs from the usual multiparty blueprint by avoiding the generation of a collective public key. Instead, each client encrypts its update under its own secret key, while the resulting ciphertexts remain compatible with homomorphic aggregation and collaborative decryption. By explicitly tracking and cancelling the ciphertext noise during decryption, the protocol removes the need for large $λ$-dependent smudging noise. We instantiate the construction with both exact BFV-based and approximate CKKS-based variants, prove its security in the semi-honest model against an adversary corrupting the aggregator and up to $L-1$ clients, and compare its communication and runtime performance with state-of-the-art MHE-based aggregation. Our results show that the proposed approach substantially reduces ciphertext expansion and online cost, while preserving practical homomorphic aggregation performance. --- ### Refining Heuristic-Based Bitcoin Address Clustering with Graph Neural Networks Authors: Hugo Schnoering, Roman Bresson, Michalis Vazirgiannis Published: 2026-09-01 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.01942v1 Bitcoin's pseudonymous nature makes it challenging to analyze user-level activity, since a single user may control multiple identifiers (addresses). Existing heuristic-based methods attempt to identify addresses belonging to the same user, but they often produce flat cluster assignments with limited modularity and are prone to errors such as merging different users together. In this work, we propose a method for refining heuristic-obtained clusters by grounding our clustering on contrastive embeddings yielded by graph neural networks. Our contributions are threefold: (i) we release a publicly available dataset of Bitcoin transaction graphs containing a substantial number of clusters; (ii) we propose a methodology for learning address embeddings consistent with heuristics, and back it up with theoretical guiding intuitions; (iii) through hierarchical clustering, we enable a finer analysis of heuristic clusters and provide a quantitative criterion for flagging suspicious merges. --- ### Sparse Readout Prism: Explaining Logit-Lens Scores in Features Instead of Tokens Authors: Matteo He, William F. Shen, Xinchi Qiu, Nicholas D. Lane Published: 2026-09-01 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2609.01936v1 A language model's prediction of its next token develops across layers, and lens methods track this process by decoding intermediate hidden states into tokens. But a lens reading reflects both the hidden state and the readout (the unembedding matrix) used to decode it. Many lenses are fit on a corpus, and we show that two lenses differing only in their fitting corpus can report different tokens for the same hidden states. We call this dependence corpus conditionality. To examine readout structure independently of the fitting corpus, we introduce Sparse Readout Prism (SRP), which decomposes the readout using only its weights and expresses any token logit or logit difference as a sum of contributions from sparse readout features. This reveals readout features as a new unit of analysis for lens readings, exposing structure that token identities can obscure and enabling comparisons across tokens, contexts, layers, and lenses. Replacing the original readout with SRP's sparse approximation reconstructs 8.9-17.3 percentage points more of the tested logit differences than the strongest of six baselines built on geometric relations among readout rows. Ablating features shifts logit differences in proportion to their SRP contributions. Although token readings vary with the fitting corpus, the dominant readout feature remains stable. Because SRP uses no corpus in its construction, it provides a control independent of the fitting corpus for lens analyses. --- ### OR-Transformer: Scaling Real-Time Decision-Making to 1,000 Items Authors: Shuze Daniel Liu, David Simchi-Levi, Claire Chen, Chutong Gao, Shangtong Zhang Published: 2026-09-01 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.01933v2 Modern supply chain operations can require coordinating replenishment across thousands of heterogeneous items under correlated stochastic demand, heterogeneous lead times, and shared fixed ordering costs, yielding observation spaces exceeding $10^4$ dimensions. At this scale, rolling-horizon stochastic mixed-integer linear programs (MILPs) become prohibitively slow, while standard reinforcement learning (RL) methods face increasingly challenging credit assignment in high-dimensional action spaces. We introduce OR-Transformer, a deep reinforcement learning framework for joint replenishment under stochastic demand, with an item-permutation-equivariant Transformer architecture and pathwise-gradient training through the inventory dynamics. Across problem sizes up to 1,024 inventory items, OR-Transformer increasingly outperforms learning-based and rolling-horizon MILP baselines as scale grows. It also reduces online decision-making time by over 4 million times relative to MILP solvers, enabling real-time, large-scale deep RL in supply chain operations. --- ### Learning with Volterra Neural Networks: A System Theoretic Perspective Authors: Haoyu Yun, Hamid Krim, Yufang Bao Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01928v2 Higher-order interaction components are important for signal, image, and video modeling, but explicit high-order operators often suffer from rapidly increasing parameter and computational costs. This paper presents kVNN, a learnable kernelized Volterra Neural operator for compact higher-order filtering. The motivation is to use kernelization to improve the efficiency of Volterra-type neural operators while providing a structured interpretation of their higher-order components. The proposed formulation combines the order-wise structure of Volterra filtering with learnable polynomial-kernel atoms, allowing different interaction orders to be represented by separate learnable centers and coefficients. This order-decoupled representation avoids explicit high-order tensor parameterization and can be implemented as a CNN-compatible layer. Experiments on representative vision tasks show that kVNN achieves a favorable accuracy--efficiency trade-off. --- ### CRISP: Cliff-awaRe Input-adaptive Sparse Prefilling with Structural-Mass-Motivated Routing Authors: Huu Huy Nguyen, Chien Van Nguyen, Franck Dernoncourt, Ryan A. Rossi, Linh Ngo Van, Jieyang Chen, Thien Huu Nguyen Published: 2026-09-01 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2609.01925v1 The attention prefilling phase of long-context LLM inference scales quadratically, making self-attention a severe computational bottleneck. Traditional sparse attention methods mitigate this through fixed patterns or offline profiling, but lack the flexibility to adapt to input-dependent attention structure. Recent dynamic methods address this by routing heads to sparse patterns in real-time, but rely on indirect routing proxies with overhead and budget allocation mechanisms that overlook the post-softmax mass hierarchy. We present CRISP (Cliff-awaRe Input-adaptive Sparse Prefilling), which identifies and addresses two structural challenges in this dynamic routing paradigm. First, we show that the routing decision can be read directly off the structure of the proxy attention map. We replace the Jensen-Shannon Divergence (JSD) routing with C_struct, a structural proxy that measures mass at Vertical-Slash compatible positions and reproduces JSD's routing decisions while eliminating both the pooled matmul and subsequent KL divergence overhead. Second, we formalize the post-softmax mass cliff and demonstrate theoretically that strictly cumulative coverage thresholds accumulate O(n) background noise at long contexts. CRISP navigates this via a sink-aware threshold grounded in the noise floor. Empirically, across InfiniteBench, RULER and LongBench on two model families, CRISP is the strongest sparse method overall and matches or exceeds exact dense attention on retrieval-heavy benchmarks, recovering up to +28.0 pp on retrieval tasks over baselines and achieving up to a 5.30x attention speedup at 512k tokens, driven primarily by our O(n) noise elimination during selection while preserving structural integrity. --- ### Looped Transformers under the Jacobian Lens: Does the Global Workspace Survive Recurrence? Authors: Wenlong Wang, Fergal Reid Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01924v1 Recent work identifies a mid-depth band of verbalisable, causally potent representations in a standard feedforward transformer --- a functional analogue of a global workspace. Whether the same workspace functionality emerges when depth is implemented through recurrence rather than a stack of distinct layers remains unknown. Looped and depth-recurrent transformers provide a direct test of this question because they reuse the same weights across depth. We extend the Jacobian lens to iterated architectures using a virtual-unrolling adapter. We apply the full workspace suite --- lens fitting, readout, and eleven causal experiment families --- to Ouro-2.6B (48 layers looped 4 times, deeply supervised) and Huginn-0125 (a 4-layer core recurred 16 times, trained for latent reasoning), using Qwen3.6-27B (64 untied layers) as the standard baseline. We find that a workspace forms in the iterated part of each architecture, but that recurrence changes how it can be accessed. Ouro reconstructs workspace content in every loop, and linear transport cannot carry that content across loop boundaries; writes and ablations must therefore span every remaining loop. Huginn carries content forward across all sixteen recurrences, while reads, writes, and ablations act only within a sliding window of roughly two recurrences. Whether newly injected content can be verbalised tracks explicit per-iteration supervision; whether existing content can be steered does not. --- ### Automated Maize Ear Phenotyping Using 3D Reconstructions Authors: Ritwesh A. Kumar, Som Tripathi, Peja Matthews, Srikar Reddy, Talukder Zaki Jubery, Patrick Schnable, Adarsh Krishnamurthy, Baskar Ganapathysubramanian Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01921v1 Maize kernel traits such as row number, kernels per row, and kernel size vary largely for genetic reasons and are consistently associated with regions of the genome that influence yield. Manual measurement of these traits, however, cannot keep pace with the volume of maize generated in a breeding program. To address this, we developed and validated a fully automated pipeline for extracting these traits from 3D point clouds of corn ears, built on a recently developed video-to-point-cloud platform. Raw video frames are processed through COLMAP and NeRF, the ear is isolated via density-based separation, and the point cloud is distance-calibrated to physical units. The calibrated ear point cloud was Z-axis aligned via PCA and cylindrically unwrapped to a 2D image. We enhanced contrast and performed zero-fine-tuning instance segmentation using Cellpose-SAM. A triple-juxtaposed unwrap strategy was used to prevent double-counting at the seam. The pipeline achieved kernel count R^2 = 0.921 (MAPE = 10.33%) and kernel row number within +-2 rows for 95.2% of ears (MAE = 0.75 rows) on a 168-ear held-out set from the 268-ear labeled dataset. The resulting multi-trait dataset has known genotype identity for each ear, positioning it for phenotype-to-genotype association analyses. --- ### Grounded, Compute-Efficient LLM Policy Agents for Energy-Poverty Equity in Physically-Constrained Peer-to-Peer Energy Markets Authors: Kunal Jadhav, Siddhesh More Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01918v1 Energy poverty is nearly absent from NLP-for-social-good, and the little existing work is either static retrieval/QA or relies on carbon-intensive cloud LLMs, a self-defeating "computational irony" for a humanitarian setting. We present EqGrid, a closed-loop simulation in which a low-frequency, open-weight LLM policy agent sets price and carbon bounds and targeted subsidies over a community of empirically-grounded household personas, while high-frequency multi-agent RL traders clear a continuous double auction constrained by a physical distribution grid (IEEE-33-bus with Dynamic Operating Envelopes). Our contribution is threefold and directly addresses how to measure the social impact of AI: (i) grounded personas (region-matched socio-demographics) whose load curves are checked for shape and level realism against real smart-meter data; (ii) formal energy-poverty equity metrics (Energy Burden, Gini of EB, LIHC) showing the intervention reduces burden inequality without raising net grid cost; and (iii) a compute-efficiency frontier that measures how much equity performance survives compressing the policy agent from a 235B teacher down to a sub-1B model deployable on a laptop, in estimated energy/carbon per decision. A decoupled-safety design (the LLM sets bounds; a validate-and-project grid gate executes) yields zero grid-constraint violations versus 55 under direct LLM control. On energy-poverty equity, the LLM policy lowers the Gini of energy burden to 0.305 (from 0.351) and mean burden by 28% while cutting cost (outperforming a tuned rule baseline), and a 3B-active model retains 95% of the benefit at roughly 9x lower inference energy than the teacher, with even a 0.8B on-device model retaining 92% at roughly 24x lower energy. We will release code and configs. --- ### Basin Geometry and Reliable Recall of Dynamical Memories in Reservoir Computing Authors: Ling-Wei Kong, Ying-Cheng Lai Published: 2026-09-01 Categories: nlin.CD, cs.LG arXiv: https://arxiv.org/abs/2609.01914v1 Reliable attractor recall conventionally requires broad basins of attraction. However, in reservoir-computing based associative memory, temporal cues reliably recover dynamical memories despite basins dominated by unpredictable, riddled-like regions. We reveal that memory basins exhibit an ``octopus-like'' structure: a robust ``head'' near the attractor and thin, intertwined ``tentacles'' spanning state space. Initial states in tentacular regions yield near-zero uncertainty exponents, making the recalled memory effectively unpredictable at finite precision. Yet, cue-driven generalized synchronization bypasses this unpredictability, driving the system into the robust basin head. This mechanism yields a quantitative relation linking minimum cue duration, synchronization rate, and basin-head radius. Trained recurrent neural networks exhibit similar geometry, suggesting this phenomenon extends beyond reservoir computing. --- ### The Ceiling Is in the Channel: Auditing Learner Gaps and Measurement Frontiers in Clinical Prediction Authors: Sayeed Shafayet Chowdhury, Nusrat Jahan, Snehasis Mukhopadhyay, Shiaofen Fang, Vijay R. Ramakrishnan Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01909v1 Clinical prediction can saturate for two different reasons: a fitted learner may fail to extract available information, or the recorded variables may impose a population frontier. We separate these quantities through the \emph{learner gap} and the \emph{measurement-channel ceiling}. Optimal balanced accuracy is characterized by total-variation separation, yielding architecture invariance, a sharp partial-identification result under replacement contamination, a cross-fitted ceiling estimator, and exact conditions for multimodal decision improvement. We add two finite-sample diagnostics, namely a label-permutation optimism floor and an underfit curve, and validate the audit on three real cohorts: UCI readmission ($n=99{,}343$), BRFSS diabetes ($n=253{,}680$), and NHANES HbA1c ($n=10{,}219$). Well-tuned gradient boosting nearly reaches the estimated frontier in UCI and BRFSS, whereas deliberately or practically deficient learners retain large gaps. NHANES yields a null difference between questionnaire and measured marginal frontiers but a significant joint complementarity gain, refining the simplistic claim that an objective modality must dominate. Across all cohorts, modest AUROC gains coexist with substantially larger Bayes decision-flip rates, and several architectures estimate similar frontiers while their achieved balanced accuracy differs sharply. A PRISMA-guided synthesis of 104 clinical tasks then shows that the same channel-level regularities recur across more than 18 disease categories: a broad but non-universal structured-clinical region, diminishing same-channel gains across model families, and higher performance when measurement channels change. The framework converts saturation from an empirical observation into an auditable decision: improve the learner when headroom remains; improve measurement when it does not. --- ### Accurate in space, unreliable in time: how LLMs represent national cultural change Authors: Yalda Daryani, Miranda Bogen, Madeleine I. G. Daepp Published: 2026-09-01 Categories: cs.CY, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.01902v1 Assessments of cultural alignment have become an important part of the development and improvement of large language models (LLMs). However, the majority of the evaluations treat culture as a single snapshot, investigating only whether a model represents a society accurately at the current time. Research in cultural psychology shows that cultural values change at different rates and directions over time. Therefore, a "culturally aware" model should capture not only where a culture is today but also how it has changed over time. We examine this missing dimension of cultural awareness using more than two decades of the World Values Survey data. We compare the cultural trajectories of 40 countries with the trajectories produced by four state-of-the-art (SOTA) LLMs on the Inglehart-Welzel cultural map. Our findings show that while models generally place countries close to their most recent surveyed positions, these representations tend to lag several years behind that position. They also capture only part of the magnitude of the observed change, introduce movement where little occurred, and rarely reproduce reversals in countries' trajectories. These findings point to temporal flattening and suggest that snapshot accuracy can give an incomplete picture of cultural awareness in LLMs and have implications for model evaluation, representational harms, and the governance of culturally aware AI systems. --- ### TAPVid-MV: A Benchmark for Tracking Any Point in 3D Across Multiple Views Authors: Skanda Koppula, Frano Rajic, Abdullah Faiz Ur Rahman, Yi Yang, Ignacio Rocco, Jeet Thakwani, Rishabh Kabra, Andrew Zisserman, Joao Carreira, Siyu Tang, Carl Doersch, Gabriel Brostow Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01899v1 Multi-camera systems are increasingly practical for robotics, AR/VR, and autonomous driving because complementary views reduce depth ambiguity and preserve visibility under occlusion. Existing point-tracking benchmarks, however, focus on a single video or static multi-camera rigs. None test long-term 3D point tracking across several synchronized views under camera motion. We introduce TAPVid-MV (Tracking Any Point in Video across Multiple Views), the first benchmark for this setting. It contains a curated set of 284 sequences, 1,142 calibrated camera streams, and 109,769 point tracks across seven subsets spanning indoor and outdoor domains, from robotics and human activity to driving and synthetic procedural scenes. We obtain these trajectories using dataset-specific auxiliary modalities: sensor depth, LiDAR, SLAM and SfM points, human meshes, posed object meshes, and simulation. Every sequence and trajectory is visually verified by human annotators. Across more than 30 baselines, no method comes close to solving the task. Surprisingly, existing multi-view point trackers do not consistently outperform monocular point trackers. By evaluating reconstruction and point tracking on the same datasets, TAPVid-MV helps distinguish errors in recovered geometry from errors in point correspondence. Through this joint analysis, we identify geometry recovery as a major bottleneck for accurate 3D point tracking. Beyond multi-view 3D point tracking, our released annotations support monocular 2D and 3D point tracking, future-trajectory prediction, and 4D reconstruction. --- ### OutageDiT: A Generative Foundation Model for Power Outage Forecasting and Scenario Simulation Authors: Yunqin Zhu, Feng Qiu, Yao Xie Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01896v1 Power-outage planning requires scenarios before an event occurs. These scenarios must represent uncertainty in magnitude, timing, and duration while preserving temporal dependence. However, severe events are rare, and data from any single region contain few examples of extreme outage and restoration patterns. To address this challenge, we introduce OutageDiT, a foundation model for generating seven-day outage trajectories at quarter-hour resolution, trained on outage and weather records across the United States. Specifically, a condition encoder processes the historical context and known future covariates once per forecast, and a shallow flow decoder reuses the resulting horizon-aligned states to generate complete trajectories. The resulting samples support point forecasting, uncertainty quantification, and conditional event simulation within one deep generative model. Across outage forecasting benchmarks, OutageDiT improves forecast accuracy and scenario quality over strong baselines and supports zero-shot transfer to held-out regions. Together, these results position conditional outage simulation as a bridge from outage forecasting to operational planning under uncertainty. --- ### Does Playing it Safe Count as Faithfulness? Reassessing LVLM Hallucination Mitigation Methods Authors: Mehrdad Fazli, Sina Mansouri, Mohit Marvania, Ziwei Zhu Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01888v1 Recent inference-time hallucination mitigation methods for large vision-language models (LVLMs) report strong gains on hallucination benchmarks. However, it remains unclear whether lower hallucination scores reflect improved multimodal grounding or more conservative generation. We evaluate six mitigation methods across three LVLMs and four benchmarks, including hallucination-focused evaluation and the diverse capability benchmark MMStar. Our analysis reveals two consistent patterns. First, hallucination reduction is often coupled with reduced informativeness: methods that lower hallucination rates also reduce object recall, visual coverage, or response detailedness. Second, improvements on hallucination benchmarks do not reliably transfer to broader multimodal capabilities, with methods showing inconsistent or degraded performance on fine-grained perception and reasoning tasks. Our findings suggest that current evaluation protocols may overestimate progress by rewarding conservative generation. We argue that hallucination mitigation should be evaluated as a faithfulness--informativeness--capability trade-off rather than through hallucination scores alone. --- ### SignMatch: Matching Dictionary Signs to Continuous Sign Language Video Authors: Ryan Wong, Youngjoon Jang, Liliane Momeni, Gül Varol, Andrew Zisserman Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01886v1 The objective of this paper is to match dictionary sign videos to corresponding signs in continuous signing videos, where a match is defined by the visual similarity alone - the handshape and motion relative to the body. To achieve this, we learn a prototype-structured sign embedding space from continuous video annotated with signs, where each learnable prototype corresponds to a sign class. Isolated dictionary videos are then mapped into this sign space, enabling the matching between dictionary exemplars and continuous sign instances. This design supports direct dictionary-guided sign matching through embedding similarity and naturally extends to unseen signs using only dictionary exemplars. Experiments on ASL-Citizen dictionary retrieval, ChaLearn OSLWL dictionary-to-continuous sign matching, and using BOBSL's CSLR2 evaluation for automatic sign annotation demonstrate strong generalisation across datasets, tasks and sign languages. Without benchmark-specific supervision, the learned representation transfers effectively across American, British, and Spanish Sign Languages, outperforming prior methods on all three benchmarks. Project page: https://www.robots.ox.ac.uk/~vgg/research/signmatch/ --- ### GAPS: Dimension-Level Gates for Conditional Activation Steering Authors: Moghis Fereidouni, Muhammad Umair Haider, Hassan Sajjad, A. B. Siddique Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01878v1 Activation steering suppresses undesired behaviors in language models by adding a steering vector to the hidden state during generation. Recent conditional methods such as CAST and DSAS improve the behavior-capability trade-off by deciding when to intervene, but once active, they apply the full dense vector to all hidden dimensions, regardless of whether a neuron carries concept information or already lies in the desired regime. We introduce dimension-level conditioning as a complementary axis of selectivity that also decides which neurons to intervene on. Our method, GAPS (Gated Activation steering via Posterior and Separability), combines two training-free gates: a static separability gate that restricts steering to neurons with statistically reliable concept information (via AUROC), and a dynamic posterior gate that steers a neuron only when its current activation is better explained by the undesired concept under a Gaussian model. The gates add O(D) overhead per token, and they plug into existing conditional methods. On toxicity mitigation (RealToxicityPrompts) and concept removal (OneSeC) with Gemma-3 (4B) and Qwen-3 (1.7B), GAPS consistently matches or improves the Pareto front of its token-level counterparts; under a fixed capability budget, DSAS+GAPS reduces Gemma-3's toxicity rate from 6.52% to 0.48%, versus 3.52% for DSAS alone. Ablations attribute most of the gain to the posterior gate. --- ### RAFT-DVC: Resolution-Aware Machine Learning-Based Digital Volume Correlation Authors: Zixiang Tong, Lehu Bu, Jin Yang Published: 2026-09-01 Categories: cs.CV, cond-mat.mtrl-sci arXiv: https://arxiv.org/abs/2609.01876v1 Digital volume correlation (DVC) provides three-dimensional full-field displacement measurements from volumetric images, but how the internal resolution of a machine-learning-based DVC model affects accuracy and operating range remains poorly understood. Here, we present RAFT-DVC, a resolution-aware family of recurrent all-pairs field transforms (RAFT)-based DVC solvers with encoder downsampling factors s = 2, 4, and 8. Using a matched design, we find that the three solvers localize displacement to approximately 0.017 feature-grid voxel, giving an empirical raw-volume error scaling of approximately 0.017s voxel. The solvers exhibit complementary operating regimes governed jointly by displacement reach and volumetric-texture compatibility. Synthetic benchmarks show that RAFT-DVC achieves errors of the same order as tuned classical DVC under fine-texture, small-to-moderate-displacement conditions and becomes competitive or advantageous under coarse-texture, large-displacement conditions. Frequency-swept tests quantify deformation spatial resolution, while tiled inference enables dense estimation on large volumes. Evaluation on confocal volumetric images acquired during indentation illustrates the importance of matching solver operating regime to deformation magnitude and image texture. Tests on micro-CT images of elastomeric foam, despite training only on particle-labeled synthetic data, provide evidence of cross-texture transfer. We also identify coordinate-order inconsistencies in three-dimensional RAFT correlation sampling and introduce a non-cubic impulse test to verify sampler geometry independently of network training. Correcting the sampler improves native-input accuracy and generalization to unseen volume dimensions. Together, these results establish RAFT-DVC as a fast, resolution-aware framework for dense DVC with characterized accuracy and operating regimes. --- ### Epistemic Sybil Resistance: Multiplying AI Agents Without Multiplying Evidence Authors: Marc Bara Published: 2026-09-01 Categories: cs.AI, cs.MA arXiv: https://arxiv.org/abs/2609.01873v1 Multi-agent AI systems improve inference by spawning agents and synthesizing reports. But another agent is not another observation: apparently independent reports may descend from the same evidence, and genuinely independent evidence can produce nearly identical reports. We formalize this as an epistemic Sybil problem. A report Z is an epistemic Sybil extension relative to reports R when I(Theta; Z | R) = 0. No report-only aggregator can generally distinguish replication from independent corroboration: identical reports can warrant different posteriors under unobserved ancestry. A Gaussian shared-root model shows common ancestry does not imply complete redundancy. Repeated extraction adds information toward a source-level ceiling, and correlated extraction errors, which a shared base model can induce among independent agents, lower that ceiling further. We test these predictions with more than 20,000 controlled LLM-agent report and extraction calls on synthetic evidentiary documents. Holding one evidence root fixed while report multiplicity rises from 1 to 32 collapses naive posterior coverage from 0.940 to 0.263. Holding report count fixed while evidence-root multiplicity rises from 1 to 16 closes the gap, and the aggregators are statistically indistinguishable at k = 16. The agent's replicate extraction errors are correlated (gamma_cal = 0.719, estimated out of sample), and a correlated-extraction aggregator restores calibration accordingly. A controlled manipulation isolates representation similarity from evidential ancestry. It changes a report-space deduplication mechanism's mean inferred cluster count by 1.425 (95% CI (1.363, 1.485)), whereas a fourfold change in true ancestry changes it by only 0.040 ((-0.045, 0.120)). Collective inference should therefore track evidential ancestry and dependence, not agent or report multiplicity or similarity. --- ### Latent unified smooth Hamiltonians for excited state chemistry Authors: David Juergens, Martin Stöhr, Andreas E. Hillers-Bendtsen, O. Jonathan Fajen, Todd J. Martínez Published: 2026-09-01 Categories: physics.chem-ph, cs.LG arXiv: https://arxiv.org/abs/2609.01871v1 We describe a neural network architecture and training procedure designed to model electronic ground and excited states of arbitrary molecular systems. By indirectly learning a latent, implicit basis representation of the electronic-state Hamiltonian, the model offers a unified treatment of multiple electronic states, conical intersections, and non-adiabatic couplings. The formalism can be further extended to learn consistent latent representations of additional operators such as transition dipole moments, for example. To demonstrate the general capabilities of our architecture, we train and evaluate networks on two realistic photochemical systems, thymine and azobenzene. The resulting models accurately reproduce energies and oscillator strengths for the ground- and low-lying excited states relevant to the photochemistry of these systems. We highlight the performance of the trained networks by studying critical molecular geometries, including conical intersections and excited state minima. By construction, the proposed framework also recovers the emergence of Berry phase accumulation around conical intersections. By pairing key mathematical structure from quantum chemistry with the representation learning power of transformers, the presented architecture offers a qualitatively new path toward fast and accurate ground- and excited-state simulations. --- ### Thinking effort aligns between humans and reasoning models in abductive reasoning Authors: Henry Arthur Published: 2026-09-01 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.01867v1 A major question in cognitive modeling concerns the behavioral alignment between large language models and humans across linguistic and non-linguistic tasks. Unlike standard LLMs, large reasoning models (LRMs) are optimized with reinforcement learning from verifiable rewards, encouraging correct solutions to reasoning tasks rather than preference-aligned responses. Recent work (de Varda et al., 2025) investigates the cost of thinking in humans and LRMs by comparing human reaction times with model reasoning traces across a range of reasoning tasks. We isolate this alignment by turning to abductive reasoning: unlike deductive tasks, its difficulty cannot be inferred from formal structure and offers no shortcuts a model could exploit to mimic effort without genuine search, providing firmer ground for empirical claims of shared effort. We find further evidence of alignment between LRM and human reasoning effort, as well as evidence that models and humans tend to make similar errors. Finally, we show that decoding methods that let models explore multiple reasoning paths increase alignment in reasoning cost between humans and LRMs across the three models tested. --- ### ExecRetrieval: Measuring the Functional-Correctness Gap in Code-Embedding Retrieval Authors: Aaryan Kapoor, Md Abdullah Al Hafiz Khan Published: 2026-09-01 Categories: cs.SE, cs.AI, cs.CL, cs.IR arXiv: https://arxiv.org/abs/2609.01865v1 Embedding-based code retrieval is a core component of coding agents and retrieval-augmented code generation, where retrieving correct code matters more than retrieving lexically similar code. Existing code-retrieval benchmarks do not plant controlled, execution-verified single-edit variants of each query's canonical implementation in the search pool, leaving the question of whether embeddings can functionally discriminate correct from near-clone-but-incorrect code unanswered in a retrieval setting. Resolving this requires a benchmark whose search pool itself contains the relevant counterfactuals -- execution-verified buggy variants near-identical to each canonical -- so that a retriever's rank ordering can be directly tested for functional discrimination rather than topical or identity overlap. We introduce ExecRetrieval, 939 Python tasks each paired with one execution-verified canonical implementation and up to four execution-verified buggy distractors, each generated by a mechanical mutation making a single targeted edit, and evaluate 23 dense embedding configurations plus BM25 under provider-native invocation with paired McNemar tests and query-level bootstrap intervals. With near-clone counterfactuals in the pool, the top hosted system reaches exec@10 = 1.00 but only exec@1 = 0.331; rank-1 misses are paired buggy variants 91.5-99.4% of the time across the four leading systems, and the canonical scores below at least one of its four paired distractors in 67-78% of queries on the leading systems. The full dataset, execution oracle, embedding matrices, environment snapshot, and pairwise statistical tests are released at the URL in Appendix D. --- ### Belief-Calibrated Optimization: An Explicit World Model for Agentic Optimization Authors: Yuhan Chen, Zhihua Tian, Mahavir Dabas, Charith Peris, Rahul Gupta, Ming Jin, Feiyang Kang, Siyuan Zhang, Nan Wang, Ruoxi Jia Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01861v1 The performance of an LLM agent depends on the scaffold around a frozen model. A common way to improve that scaffold is to use a coding agent as an optimizer: it reads current scores and traces and iteratively edits the source, producing a new candidate each round. Each edit is chosen according to a belief about how the environment will respond: what went wrong, and which change should help. That belief is typically implicit. It lives in the coding agent's reasoning on the current call, or remains latent in its parameters, rather than as something written down. Later calls therefore see scores and traces, but they do not use that belief. We introduce Belief-Calibrated Optimization (BCO), a method that writes that belief down as a persistent in-context document and continually revises that document as new candidates are evaluated. The resulting document is a world model: the current account of how the environment responds to edits. Added to an otherwise standard loop, BCO reaches a higher train passrate than a matched control that lacks only the world model, on five benchmarks spanning memory QA, tool-use QA, code-as-action app agents, and terminal agents. The gap remains on every held-out split, which is not used to select the candidate. After a target-model swap, in which the frozen model is replaced and the scaffold is not, the selected BCO scaffold leads on the tasks we test, except where context-window overruns leave it unfinished. An offline ablation then asks whether that gap comes from what the world model says. A fresh predictor given the accumulated document forecasts how the environment will respond more accurately than predictors given either no document or a same-form copy whose content has been falsified. The comparison indicates that the document carries reusable information in its content, not only in its form. --- ### The Memory Trust Gap: Capability-Dependent Failures in Persistent-Memory Agents Authors: Jundong Hu, Shekar Ramachandran Published: 2026-09-01 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2609.01852v1 Persistent memory supports personalized agents, but a stale stored fact can override current authoritative evidence without warning. We study when this harm begins as model capability changes. We evaluate a frozen, closed-set, action-scored benchmark with 2 suites that represent 2 different meanings of "no memory" (a Benefit suite, unsolvable without the stored fact, and a Safety suite, in which an authoritative tool always holds the correct value), on a same-family model-size series (Qwen3 0.6/1.7/4/8B). The Memory Trust Gap reflects over-trust rather than confusion. In the Benefit suite, models answer with the stale value 0.92-1.00 of the time at every scale. In the Safety suite, harm below the no-memory baseline under the trap conditions ($Δ_{\mathrm{mem}}$) is capability-gated, with the larger models collapsing most once a stale note is made to look current. In a $2\times2\times2\times2$ factorial, which feature triggers over-trust depends on both the feature and model scale. Removing a label amplifies over-trust at every size, and a recency feature (stale dated newer) fools the larger models harder. Source authority is weak and scale-flat, and position changes from positive to negative across the Qwen3 model-size series. We confirm these scale interactions with direct cross-size contrast tests rather than overlapping per-model intervals. Mitigation is likewise capability-dependent: exposing metadata improves accuracy for the capable models, but only pre-resolving the conflict restores accuracy for the 2 smaller checkpoints. The same pattern appears on the capable models in an independent Llama-Instruct model-size series and on 2 external datasets (RGB, MisBench). A framing control finds no consistent advantage for the memory label: at the 3 smaller scales, models trust a stale document more than a stale memory; at 8B, the difference is not significant. --- ### SSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval Authors: Przemysław Stokłosa, Janusz A. Starzyk, Paweł Raif Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01849v1 This article presents SSAKG 2.0, an open-source software package for constructing and operating Structural Sequential Associative Knowledge Graphs (SSAKGs). An SSAKG represents objects as graph vertices and ordered sequences as structural patterns of graph connections. The resulting sparse graph is used as an associative memory in which complete sequences can be reconstructed from a partial, unordered context. Version 2.0 introduces new algorithms that exploit individual bits of computer memory to efficiently search graph connections. The package is implemented in Python, while performance-critical graph operations are implemented in C and exposed through a Python interface. This hybrid implementation provides a flexible high-level programming environment while reducing the memory and computational overhead associated with large sparse graphs. The algorithms were evaluated using randomly generated numerical sequences, sequences derived from sentences in the NLTK corpus, and mRNA sequences. The experiments demonstrate the ability of the package to store and reconstruct sequences from partial contexts and provide a basis for evaluating the effects of graph density, sequence length, and memory size on retrieval performance. SSAKG 2.0 is distributed under the Apache 2.0 open-source license. The package includes documentation and reproducible examples and is publicly available through GitHub and the Python Package Index (PyPI). --- ### Cite or Decline: A Strict Course-Grounded Chatbot for STEM Lecture Videos Authors: S M Masrur Ahmed, Jaspal Subhlok Published: 2026-09-01 Categories: cs.CL, cs.CY, cs.IR arXiv: https://arxiv.org/abs/2609.01846v1 Recorded lecture videos, often enhanced with search and summarization features, are a standard study resource. However, students cannot easily ask course specific questions or verify answers against an instructor's lecture. We report a semester-long deployment of VideoPoints platform with a retrieval-augmented chatbot that answers from course lecture materials and returns timestamped citations. The chatbot retrieves only from the active course, uses chapter summaries to guide transcript ranking, and returns clickable timestamped citations. Students used it for quick lookups and exam review. Across 833 messages, 70.5% included citations, none crossed a course boundary, and when no lecture evidence matched, the chatbot usually declined rather than answering. Among the users, citations were the most consistently useful feature, while practice-question generation was the strongest unmet request. We also evaluated the design on the real-world test split of EduVidQA, a public multimodal benchmark for lecture-video question answering. Our design improved correct-lecture retrieval by 6.3 percentage points over dense-only retrieval. Together, the results show that effective deployment depends on course isolation, supported citations, and alignment with students' study practices. --- ### Cross-Model Distillation of a Human-Pose Foundation Model from Unannotated Infant Video for Markerless 3D Pose Estimation Authors: R. James Cotton, Divya Joshi, Colleen Peyton Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01840v1 Spontaneous movement is one of the earliest windows onto an infant's neuromotor health, and structured clinical instruments that score it are validated early predictors of cerebral-palsy risk. However, they require specially trained raters, are time-consuming, and carry inter-rater variability. This motivates automated, video-based markerless assessment, especially as marker-based motion capture is impractical in infants. Yet the foundation models that make markerless capture possible are trained almost entirely on adults: our recent multi-view infant study found that no single model is jointly best, with strong 2D keypoint accuracy and direct 3D body recovery split across different models. While that study identifies this trade-off, it does not resolve it. Here, we perform cross-model distillation from the Sapiens 2 pose model into the SAM 3D Body model, using unannotated infant video alone. A frozen teacher supplies dense pseudo-labels, and a differentiable renderer aligns the predicted mesh to them in the training loop. On eleven held-out infants (18 sessions, 173 recordings) under our prior study's multi-view protocol, fine-tuning improves same-view 2D keypoint agreement with the Sapiens reference (median body percentage of correct keypoints @ 10px 0.22 -> 0.42, face 0.22 -> 0.42) and Procrustes-aligned mean per joint 3D position error (25.5 -> 22.2 mm). This demonstrates how cross-model distillation improves SAM 3D Body model performance on infants. --- ### Import What You Need: Learning When and How to Augment EHR Graphs with External Knowledge Authors: Chen Chen, Mohsen Nayebi Kerdabadi, Dongjie Wang, Mei Liu, Zijun Yao Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01839v1 Longitudinal prediction from electronic health records (EHRs) is limited by the sparsity and irregularity in patient trajectories, and knowledge augmentation with external knowledge graphs (KGs) offers a promising way to alleviate these issues. However, most existing methods perform fixed, context-agnostic topology augmentation by adding the same KG nodes and edges regardless of a patient's evolving state. We propose ReTA, a Reinforcement learning-based dynamic Topology Augmentation framework that casts KG import as a per-visit, budget-aware policy. ReTA first constructs an offline refined pool of KG-grounded templates, then learns a policy to select one augment action per visit from three options: Soft Import, which enriches node features without modifying graph topology, Hard Import, which grafts a compact KG subgraph onto the visit graph to create message-passing shortcuts, and Skip, which leaves the visit unaugmented when the base encoder is already confident. To stabilize learning, ReTA employs a decoupled encoder that processes semantic and structural signals in separate channels and fuses them via adaptive gating. Experiments on MIMIC-III and MIMIC-IV across diagnosis prediction, mortality, and readmission show that ReTA consistently outperforms strong baselines while remaining efficient, transfers across datasets and knowledge graphs, and yields interpretable augmentation patterns. The robust gains under sparse supervision highlight the advantage of ReTA's dynamic decision to import knowledge, boosting accuracy while curbing costs. --- ### Agent Memory Is a Surface for Endogenous Authorization Laundering Authors: Tommaso Cerruti, Mika Okamoto, Ansel Kaplan Erol Published: 2026-09-01 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2609.01836v1 Long-running LLM agents rely on persistent memory to carry state across interactions, including permissions, restrictions, and revocations. When memory misrepresents this evolving authorization state, the agent's own records can grant authority that the underlying history never permitted, resulting in misaligned behavior without any external attacks. We term this failure endogenous authorization laundering, where spurious permissions written into memory lead to unauthorized actions as their provenance is washed away. We then introduce EAL-Bench, which measures how accurately persistent memory preserves evolving authorization state and whether errors propagate to downstream unauthorized actions. We evaluate five LLMs as memory writers and two as executors across procurement, cybersecurity, and finance. We find that under incremental memory updates, writers create false authority for up to 50.2% of unauthorized requests; once false authority is present, executors act on it in 98.6% of trials. Two safeguards, requiring stored permissions to be backed by valid source events, and tracking permission changes through bounded event sourcing, substantially reduce laundering, but both also reject more legitimate actions, exposing a safety-utility tradeoff. Persistent memory is therefore not merely a performance component, but a part of an LLM agent's effective authorization policy. --- ### Architecting Conversational Data Systems for Stateless LLM APIs: The Hydration Proxy Pattern Authors: Joseph Axisa Published: 2026-09-01 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2609.01834v1 As enterprise platforms transition to conversational reasoning interfaces, the stateless nature of LLM APIs creates an architectural gap. While statelessness enables horizontal scalability for AI providers, it forces client applications to manage the entire burden of conversational state and semantic memory. The work identifies the Hydration Proxy Pattern, an architecture that decouples session persistence from the reasoning engine. The framework ensures platform sovereignty over conversational data while enabling secure, multi-stage semantic grounding. We further propose the Context Stabilization Mandate to resolve the tradeoff between sovereign state management and KV caching. --- ### Candidate Generation and Definition-Guided Verification for Sentence-Level Depression Symptom Recognition Authors: Weiming Li, Catarina Barata, Miguel Constante, Joao Sanches Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01833v1 Sentence-level recognition of depression symptoms is challenging because similar expressions can differ in symptom relevance, and language-model inference is insufficiently grounded in diagnostic definitions. This study proposes a two-stage framework separating symptom-candidate generation from definition-grounded verification. A contrastively fine-tuned sentence encoder generates a symptom candidate per sentence, and a fine-tuned language model verifies whether the candidate is present or absent using the sentence, its context, and a candidate-specific diagnostic definition, checking its judgment against that definition before answering. Evaluated against encoder, inference-based, medical, and general LLM baselines and a matched single-stage supervised classifier, the proposed pipeline attains the best accuracy and F1 scores of all methods, with rationales matching expert-authored annotations. A preliminary clinical audit indicates moderate alignment with diagnostic definitions, with explanation quality strongly dependent on prediction correctness. The results support decomposing symptom recognition into candidate generation and definition-grounded verification, though performance remains limited for rare categories. --- ### Interpretable Symptom Vectors for Depression in a Large Language Model Authors: Fangyi Zhu, Ajay Subramanian, Allison Constant, Camille Wang, Ravish Gupta, Corey J. Keller Published: 2026-09-01 Categories: cs.CL, cs.AI, cs.LG, q-bio.NC arXiv: https://arxiv.org/abs/2609.01832v1 Patients with depression present with diverse symptom profiles, yet clinical practice routinely reduces this variation to a single severity score. Large language models (LLMs) can potentially capture various symptoms and their severity from patient speech. However, how depressive symptoms are represented inside LLMs remains poorly understood, limiting clinical trust. To examine whether internal model activations match clinician judgment, we analyzed the residual stream of Gemma-3-27B-PT using mechanistic interpretability techniques. Recording activations across symptom descriptions drawn from validated clinical instruments, we found that symptom groups geometrically separated the most at layer 21 across multiple distance metrics. Using Semantic Projection, we then projected held-out naturalistic text onto Symptom Vectors constructed from these instruments. The resulting per-symptom coefficients preserved clinician-annotated rank ordering across mood, somatic, and suicidality axes. Furthermore, a single depression vector in Layer 21 separates held-out depressive from non-depressive text (AUC = 0.789), which can be used as an emotional valence gate that restricts symptom projection to depressive speech. These results reveal a decorrelated, clinician-aligned symptom signal readable directly from internal activations, offering a mechanistic foundation for interpretable depression-assessment tools. --- ### AVERT: Audio-Verified Adjudication for Spoken Dialogue State Tracking Authors: Chunggi Lee, Hanspeter Pfister Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01828v1 Spoken dialogue state tracking recovers slot-value pairs from speech, where ASR errors concentrate in entity values and persist across turns, making it both a generation and an editing problem. A strong per-turn text editor corrects much of this but, operating on the transcript alone, leaves three recoverable errors: a value predicted inconsistently across turns, an omitted slot, and a value the audio does not support. We present AVERT, which scores each candidate value by combining cross-turn agreement with a trained audio-conditioned verifier and resolves the three error types with three operators, vote, add, and swap, each restricted to the slots where its error is common. On SpokenWOZ, a base speech-LLM reaches 33.04 JGA, a text editor 38.34, and AVERT 40.13, without retraining either. This is in the range of a 1B end-to-end system that consumes the full spoken history (39.32), though AVERT uses two 1B decoders rather than one. The audio verifier contributes a statistically significant gain, and restricting each operator to a selected slot subset matters: removing it lets unrestricted voting overwrite correct categorical values and fall below the editor. ---