# Research Papers — 2026-08 · bundle page 61 of 240 > Full-text markdown of papers #3001–3050 of 11995 published in 2026-08, newest first. AI-readable mirrors of the full 11995-paper 2026-08 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-08/page/61 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/60.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/62.md) ### Calibration-Preserving Pruning: Compression as a Reliability Contract Authors: Ibne Farabi Shihab, Adria Binte Habib, Anuj Sharma Published: 2026-08-24 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2608.23744v1 Split conformal prediction, not the pruning rule, supplies finite-sample marginal coverage once a pruned model is fixed independently of the conformal calibration split. We study the separate efficiency problem: can pruning preserve score geometry well enough to obtain smaller valid prediction sets? Calibration-Preserving Pruning (CPP) augments a base pruning score with nonconformity-gradient saliency and uses disjoint pruning, validation-selection, conformal-calibration, and test splits. Bounded score perturbations imply bounded conformal-quantile shifts and controlled set inflation, but do not make the generic coverage theorem CPP-specific. Final five-seed Qwen2.5-1.5B results at 50\% sparsity show the largest gains on large-label tasks. On DBpedia-14, CPP-SparseGPT reduces mean set size from \(10.1\) to \(8.6\) while changing accuracy from \(0.347\) to \(0.366\); CPP-Wanda reduces \(11.2\) to \(9.0\) with an accuracy trade-off from \(0.310\) to \(0.295\). Across 15 dataset--sparsity cells, CPP-SparseGPT produces smaller sets in 13 and higher accuracy in 11. Matched controls show that generic supervised gradients explain much of the gain: true-label CPP is not statistically resolved from matched Wanda+SNIP, whereas threshold-aware candidate-label CPP reaches \(7.8\) mean set size at explicit accuracy and offline-compute costs. RoBERTa-base and Llama-3-8B diagnostics support transfer, but our claims remain limited to reliability-sensitive classification. --- ### AgentRoom: Concurrent Multi-Agent Coding in a CRDT-Backed Shared Workspace Authors: Seonglae Cho, Donghyun Lee Published: 2026-08-24 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2608.23740v1 Concurrent multi-agent coding promises division of labor across modules, robustness through redundancy, and parallel exploration at the natural granularity of multi-file projects. Realtime collaborative editing protocols solve this coordination problem for human teams via Conflict-free Replicated Data Types (CRDTs), but the LLMs underneath generate one token at a time and existing multi-agent coding systems inherit this serial limit: they either sequence agents through phase handoffs or pool independent samples without coordination, and a single agent abandons up to half of hard tasks with a one-file stub-and-exit. AgentRoom is a realtime collaborative editing protocol for concurrent coding agents. Its runtime layer exposes file-level claim, status, and broadcast as MCP tools on a CRDT-merged shared filesystem. Five frontier coding-CLI models ran four backend coding tasks, with cross-language checks in Python DevBench and Rust+axum. For CLI-stable models, AgentRoom with 2 agents abandons fewer tasks than Solo and has less run-to-run variation. At matched-compute, one positive mean LLM-judge contrast puts AgentRoom over parallel-merge. The other contrast, a bundle probe, puts full AgentRoom above each partial case: an ordering rather than a percentage split. Coordination, not parallelism or CRDT-merge, bears the load. --- ### More Motion Is Not Always Better Motion: Corpus Composition Governs Whether Augmentation Helps SMPL-Based Parkinsonian Gait Severity Estimation Authors: Michael Caiola, Andrew C. Weitz Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.23730v2 We grade MDS-UPDRS gait severity from SMPL motion using three frozen MotionAGFormer encoders as featurizers, reaching macro-F1 0.58 on a hidden, multi-site test set. Because the system's members differ only in their lifting corpus, evaluating encoders singly on that test set isolates what that corpus contributes. Six pools drawn from one inertial dataset, varying only in which walking tasks they include, score between 0.32 and 0.53, and just one of them beats the 0.51 of an encoder given no outside motion at all. What separates them is not how much data they hold but whether they carry a contrast in walking speed, the variation this representation appears to depend : a further pool adding a third collection site at fixed task composition does worse still. The same rule explains why exact synthetic motion and monocularly reconstructed web video both fail to help. Modifying the learned representation itself, rather than the corpus behind it, cost every variant that attempted it. --- ### Velocity-coupled Representation Refinement for Satellite Orbit Prediction Authors: Yue Yang, Zhiqiang Wu, Saiyu Qi, Fan Ma Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.23728v1 Satellite orbit prediction, which aims to forecast future orbital trajectories from historical observations, is important for collision warning and safe space operations. With advances in time-series forecasting, learning-based methods have emerged as a promising solution for satellite prediction. In orbital dynamics, a satellite state is typically described by position and velocity, where position characterizes trajectory geometry and velocity reflects its instantaneous direction and rate of change. However, most existing methods mainly focus on temporal dependencies within position sequences while rarely exploiting the intrinsic coupling between position and velocity, which is essential for modeling satellite motion. To this end, we propose OrbitNet, a velocity-aware representation learning method for accurate satellite orbit prediction. It lifts conventional position-sequence forecasting to a position-velocity coupled representation learning paradigm by exploiting relationships among satellite state variables. Specifically, we develop a velocity-coupled representation refinement strategy to enhance positional representations through cross-variable interactions between position and velocity. We further introduce orbital segment modeling, which partitions historical trajectories into temporal segments and performs segment-level temporal learning to capture local motion variations and long-range evolution patterns. Extensive experiments show that OrbitNet outperforms large time-series foundation models and representative general forecasting methods under both in-domain evaluation on Starlink and zero-shot evaluation across six unseen satellite constellations. We expect this work to encourage further exploration of satellite-aware representation learning for trajectory time-series forecasting. --- ### A Hybrid Two-Stage Machine Learning Pipeline for Fault Detection and Classification in Power Transmission Systems Authors: Sahil Manikshete, Atharva Gujarathi, Thanh Long Vu, Akhtar Hussain, Van-Hai Bui Published: 2026-08-24 Categories: eess.SY, cs.LG arXiv: https://arxiv.org/abs/2608.23726v1 Rapid and accurate fault detection in high-voltage transmission networks is essential for grid reliability and equipment protection. Transmission fault datasets are frequently imbalanced, and certain fault types produce electrical signatures that fall within the normal operating envelope, causing single-model classifiers to fail on safety-critical cases. This paper proposes a hybrid two-stage machine learning pipeline that decouples detection from classification. Stage 1 combines an Isolation Forest anomaly detector with an optional supervised binary detector through an OR-fusion rule; the supervised branch is allocated automatically during training for any fault class the anomaly detector cannot resolve, and is omitted when no such class exists. Stage 2 applies a Random Forest multiclass classifier only to samples flagged by Stage 1. Feature engineering is expressed as a per-measurement-point operator mapping six raw channels to eighteen features, including zero-sequence symmetrical components derived from Fortescue's theorem, yielding 18L features for L measurement points. On the TLFaultDataset, the pipeline raises Line-fault end-to-end accuracy from 31.3% to 95.8%. On an independent single-point dataset, the same framework attains 97.25% end-to-end accuracy across all classes including normal operation, exceeding the TLFed federated benchmark of 94.84% without GPU or federated infrastructure, at 0.05 ms per sample on CPU. Ablation on both datasets shows zero-sequence features resolving the three-phase versus three-phase-to-ground ambiguity, raising the F1-score of that class pair from 0.39 to 0.997. The direction of the zero-sequence signature is found to be system-dependent, motivating a learned decision boundary in place of a fixed relay threshold. --- ### Response Renormalization for Critical Deep Equilibrium Models Authors: Jose Luis Lima de Jesus Silva Published: 2026-08-24 Categories: cs.LG, cs.NE arXiv: https://arxiv.org/abs/2608.23725v1 Deep Equilibrium Models (DEQs) compute predictions from a hidden representation unchanged by the model update. Training through this equilibrium uses implicit differentiation and requires solving an adjoint system built from the residual Jacobian. If this Jacobian is nearly singular along loss-sensitive directions, small perturbations can be strongly amplified in the adjoint response, producing large, highly sensitive gradients that can make optimization unreliable. We introduce Response Renormalization, a backward-pass framework that lifts selected near-pole denominators while leaving unlifted response channels unchanged. Collective Mode Response Renormalization (CMR) applies this correction in a low-dimensional critical subspace, while Phi-adaptive CMR computes a bounded response mass from a positive susceptibility rule. We derive dense and matrix-free collective formulations, distinguish exact gradients of a modified frozen-anchor residual from backward-response surrogates, and extend the construction to Structured Implicit Layers and Vector Attractors (SILVA). Across 23 multiphysics families spanning partial differential equations, three-dimensional fields, operator maps, complex geometries, and particle systems, CMR and Phi-CMR yield test errors no more than five percent higher than those from models trained with exact implicit differentiation in more than 98% of static and 95% of transient family-seed comparisons. Solver-index experiments show convergence toward the static adjoint, while physical-time rollouts retain predictive fidelity under the evaluated conditions. These results demonstrate that selective response renormalization can control near-critical adjoint amplification without globally damping well-conditioned sensitivity. Therefore, the method can make parameter updates more reliable while preserving the useful gradient information needed for learning. --- ### DriftAD: Visually-Guided Text Drift for Few-Shot Industrial Anomaly Detection Authors: Wenyang Liu, Tianyi Liu, Dongshuo Zhang, Kejun Wu, Adams Wai-Kin Kong Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.23723v1 Few-shot anomaly detection (FSAD) has recently benefited from vision-language models such as CLIP, which enable anomaly de?tection by aligning visual features with text descriptions of normal and abnormal states. However, existing methods typically rely on static text prompts that are applied uniformly across the entire feature hierarchy and spatial dimensions. This rigid global-to-local matching fails to capture the highly localized and scale-dependent physical variations of industrial defects. To address this, we propose DriftAD, a FSAD framework built on three key modules. First, an Anomaly Signal Amplification (ASA) module enhances subtle defect signals through spatial and frequency branches before text-visual matching. Second, Visually-Guided Text Drift (VGTD) dynamically transforms frozen CLIP text embeddings, steering them into layer?wise, spatially-adaptive anomaly descriptors conditioned on local visual context at each encoder depth. Third, Drift-Guided Spatial Gating (DGSG) uses the drifted abnormal descriptor as a spatial probe to selectively enhance anomaly-relevant visual features. Addi?tionally, a drift separation loss prevents representational collapse of the drifted descriptors, and a gate supervision loss enforces spatially discriminative gating in DGSG. Extensive experiments on MVTec?AD and VisA demonstrate state-of-the-art performance across all 1-, 2-, and 4-shot settings on both image-level and pixel-level metrics. Code is available at https://github.com/wenyang001/DriftAD. --- ### Platonic Representation Hypothesis on World Models Authors: Wenhow Li, Chengwei MA, Hui Xiong, Ying-Cong Chen, Lei Zhang Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.23720v2 World models have demonstrated significant potential for perceiving and simulating complex environments. Despite their strong performance, the fundamental nature of their learned representations remains poorly understood. In this paper, we investigate the Platonic Representation Hypothesis within this domain by proposing the Predictive Consistency Assumption: we posit that the optimization of a shared state transition objective acts as a selective pressure that encourages heterogeneous models to converge toward a shared latent structure. Through systematic experiments with the DINO World Model (DINO-WM), in which we vary visual encoders to create heterogeneous models, we find that capable world models evolve toward geometrically similar internal structures. Moreover, via model stitching, we show that the internal features of one world model can be mapped to another with limited performance degradation, providing evidence of functional compatibility. Our findings suggest that the pursuit of predictive consistency can promote shared, transition-compatible latent structure across world models. --- ### ADE: Agentic Data Evolution Framework for Human-Centered Objectives Authors: Yang Yu, Yilin Jiang, Zexuan Fei, Yiming Luo, Xingkai Song, Kaiyi Huang, Aimin Zhou, Xin Lin, Fei Tan Published: 2026-08-24 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.23719v1 Aligning large language models to human-centered objectives is difficult when targets are non-executable and context-dependent, limiting reliable verification and scalable supervision. Although synthetic data expands coverage, weak verification shifts the bottleneck from generation to selection. Noisy signals destabilize iterative refinement and can cause silent regressions. We propose Agentic Data Evolution (ADE), a data-centric framework that organizes synthetic supervision as evolving data snapshots. ADE improves data snapshots through a closed-loop Observation-Variation-Selection (OVS) procedure, where a steady-state admission mechanism acts as a quality ratchet that conservatively gates updates for sustained cross-round improvement. We validate these improvements through complementary intrinsic trend tracking and extrinsic post-training evaluation. On DEV300, ADE raises the intrinsic win rate from 50% to 75.81% and the extrinsic win rate from 55.20% to 68.86%, consistent performance gains across diverse benchmarks. Blind expert evaluation further confirms this, with a 66.11% preference for evolved answers. These gains extend across post-training methods, model scales, and tasks beyond the target weakly verifiable educational objectives. Resources are available at https://github.com/ZeroLoss-Lab/Agentic-Data-Evolution. --- ### Integer Natural Evolution Strategies Authors: Jacob de Nobel, Diederick Vermetten, Hao Wang, Ofer M. Shir, Michael Emmerich, Thomas Bäck Published: 2026-08-24 Categories: cs.NE arXiv: https://arxiv.org/abs/2608.23714v1 While contemporary Evolution Strategies handle integer optimization problems effectively, their adaptation mechanism is grounded in $\ell_2$-based Gaussian models, which are not native to the integer lattice. In contrast, the $\ell_1$-norm provides the natural measure of displacement on $\mathbb{Z}^n$, with the double geometric distribution as its canonical mutation operator. In this work, we derive a fully $\ell_1$-native step-size adaptation mechanism from first principles and propose an Integer Natural Evolution Strategy. We show that the DG distribution belongs to the exponential family, and that its sufficient statistic $|z|$ yields a natural-gradient signal for dispersion adaptation. By accumulating this signal via an evolution path, we obtain a fading-memory online estimator of the natural gradient, following Ollivier (2018). This establishes that DG-based step-size adaptation arises directly from the statistical structure of the mutation distribution, rather than as a discrete analog of continuous ES mechanisms. Empirical results on integer quadratic benchmarks show that \textsc{INES} learns meaningful coordinate-wise step-sizes and is competitive with integer-handling CMA-ES baselines. Its advantages are most visible in high-dimensional Ellipsoidal problems and in robust convergence at larger dimensions. --- ### Do LLMs Understand Limit Order Book Dynamics? Authors: Junxiao Chen, Paul Glasserman Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23706v1 A large language model (LLM) trained on synthetic limit order book (LOB) data achieves near perfect scores in generating valid sequences of LOB events. However, the LLM's implicit world model fails to learn the state of the LOB. This deficiency leads to biased estimates and spurious predictability in using the LLM to forecast future LOB events. Our analysis uses novel tests of an LLM's world model, extending prior work from deterministic settings to the stochastic dynamics needed for the LOB. --- ### The Limits of Automatic Evaluation of Creativity in Large Language Models Authors: Alessandro Tutone, Giorgio Franceschelli, Mirco Musolesi Published: 2026-08-24 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2608.23705v2 Large Language Models (LLMs) are increasingly capable of generating text that challenges human performance in domains requiring creativity, yet evaluating creativity in LLM-generated content remains a significant challenge. Here, we investigate whether current automatic evaluation methods can reliably capture human judgments of creativity. We collect human evaluations of human- and AI-generated short stories from the WritingPrompts dataset across 11 dimensions of creativity, and compare these judgments with automated objective metrics and LLM-as-a-Judge evaluations. Our experiments reveal substantial misalignment between automatic evaluations and human assessments. In particular, LLM-based judges exhibit a systematic preference for AI-generated stories, consistently favoring their stylistic characteristics over the unpredictability and other qualities of human-authored texts. Furthermore, correlation analyses show that widely used automatic metrics exhibit near-zero alignment with human judgments across both human- and AI-generated stories, suggesting that they fail to capture important dimensions of creativity. These findings highlight fundamental limitations in current approaches to the automatic evaluation of creative text and underscore the difficulty of reducing the multidimensional and subjective nature of creativity to computational metrics. --- ### Renormalization Group Flow Matching for Scalable Local Generative Modeling Authors: Kanta Masuki, Yuto Ashida Published: 2026-08-24 Categories: cs.LG, cond-mat.stat-mech arXiv: https://arxiv.org/abs/2608.23696v1 Despite their remarkable success in modeling complex data, generative models face a fundamental tradeoff. Global approaches can capture full structural coherence but suffer from high computational costs, while local models are efficient but often fail to reproduce long-range correlations and global coherence. The renormalization group (RG) bridges this gap by seamlessly connecting spatial structures across different length scales, retaining quasi-local descriptions at each step while preserving long-range correlations. We introduce renormalization group flow matching (RGFM), a generative framework that systematically structures data generation across different spatial scales. By using an exact RG flow as the probability path, RGFM progressively generates data from long- to short-wavelength structures. To reconcile scalability with global structure, we exploit two key properties of the RG: quasi-locality and scale separation. We rigorously show that the RGFM probability flow can be accurately approximated by local velocity fields acting over a spatial range $O(Λ^{-1}(\ln L+\ln(1/\varepsilon)))$ for RG wavenumber scale $Λ$, linear system size $L$, and prescribed error tolerance $\varepsilon$. This property enables local generative modeling with patches of size $O(\ln L)$ and a computational cost that scales nearly linearly with the system volume. We numerically demonstrate that local RGFM reproduces long-range correlations far beyond its receptive field in representative one-dimensional distributions, while conventional local flow matching exhibits substantial errors at long distances. On FFHQ images, RGFM yields far more coherent and higher-quality samples than local flow matching at 64x64 and 256x256. Our results establish RG-guided probability flows as a promising route toward scalable generative modeling that captures long-range structure using only local computation. --- ### Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment Authors: Stephen Chung, Wenyu Du, William J. Wesley Published: 2026-08-24 Categories: cs.AI, cs.DM, cs.MA arXiv: https://arxiv.org/abs/2608.23691v2 We study autonomous mathematical discovery in the Station, an open-world multi-agent environment in which AI agents from different model families pursue a shared research goal without a central coordinator or scripted pipeline. Agents choose their own research directions, conduct experiments, collaborate and publish papers. These papers accumulate into a shared body of knowledge that later agents can read, cite and extend. We evaluated the Station on 12 mathematical construction problems from the AlphaEvolve study and two additional case studies. Five of the 12 problems yielded results novel relative to the prior literature: a new infinite family of finite field Kakeya sets, new exact 604-point kissing configurations in eleven dimensions, improved bounds for the discretized Kakeya needle and sign uncertainty problems, and a substantially improved lower bound for Erdős's minimum overlap problem. Agents also discovered novel infinite families for Book Ramsey numbers. Their research extended beyond searching for high-scoring constructions: agents developed explanations of their findings and proved theorems outside the assigned tasks. These explanations guided further discoveries and were preserved in the agents' papers, making the underlying insights easier for external researchers to understand and build upon. All presented discoveries are supported by exact constructions or proofs formally verified in Lean. We release the source code, full agent dialogues, papers and verification code, providing a transparent record of how these discoveries emerged. --- ### Automata from Agent Traces: Failure and Next-Step Prediction Authors: Seonglae Cho, Franklin Cardenoso Fernandez, Umar Mohammed, Zekun Wu, Kleyton Da Costa, Ilham Wicaksono, Adriano Koshiyama Published: 2026-08-24 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.23670v1 LLM-based agents execute multi-step tasks, but their behavioral structure remains opaque: long unstructured traces resist the safety auditing and runtime monitoring that deployment requires. Existing approaches operate per-trace or success-only, so they miss the cross-run topology that links next-step and failure prediction. To recover that shared structure, we collapse an entire trace corpus into a single, compact finite-state machine (FSM) that serves as a structural substrate for the otherwise unpredictable behavior of LLM agents. Across twelve public datasets, the FSMs are compact (7-43 states), replay held-out data at >=0.997 fitness with near-identical topology across splits, and build in milliseconds. This substrate addresses both prediction goals. For next-step prediction, FSM-state context outperforms Agent Workflow Memory on every ground-truth-matched dataset. For failure prediction, per-state behavioral features reach held-out AUROC up to 0.94, and an online monitor ranks failing runs above passing ones from a partial trace, triggering early stopping well before completion. Behavioral topology thus appears shaped more by the deployment harness than by the LLM, providing a model-agnostic structural primitive for safety auditing and runtime monitoring. --- ### Gated Activation Steering for Reducing Sycophancy & Hallucination in Medical Question Answering Authors: Himanshu Tripathi, Subash Neupane, Shaswata Mitra, Sudip Mittal, Noorbakhsh Amiri Golilarz, Shahram Rahimi Published: 2026-08-24 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.23666v1 Sycophancy and hallucination are persistent failure modes of Large Language Models (LLMs) across domains. However, it becomes particularly consequential in clinical question answering, where responses must remain grounded in the provided context and robust to user pressure. Hallucination can introduce information that is unsupported by the context, while sycophancy can cause a model to abandon a previously correct answer when challenged by the user. Existing approaches, such as prompt-based safeguards and always-on activation steering, often address these behaviors separately or apply interventions broadly across turns, which can unnecessarily deteriorate responses that were already correct. To address these limitations within a single framework, we employ Inference Time Intervention (ITI) to jointly control both behaviors by learning separate steering directions for hallucination and sycophancy from contrastive clinical pairs and applying them to causally verified attention heads. During runtime, behavior-specific gates then determine when intervention is needed: the hallucination component mitigates unsupported claims, while the sycophancy component mitigates answer shifts caused by user pressure. We evaluate this framework on clinical questions grounded in EHR data while keeping the model weights frozen. Across all evaluation settings, we conducted 15,900 model-response runs. Across 600 pressure trajectories for the 4-billion-parameter model, the unsteered model caved in 570 cases. At the same time, gated steering helped it last longer in 551 of them. It held its ground under pressure at levels comparable to those of models with more than 100 billion parameters, showing that targeted inference-time steering can improve robustness without intervening at every turn. --- ### Scaling Reinforcement Learning for Diffusion Models via Velocity Matching Authors: Jaemoo Choi, Wei Guo, Yuchen Zhu, Arash Vahdat, Molei Tao, Julius Berner, Yongxin Chen Published: 2026-08-24 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.23664v2 Reward-based fine-tuning of diffusion models has largely inherited likelihood-based policy optimization developed for autoregressive large language models (LLMs). Diffusion models, however, are natively trained through velocity regression and do not directly provide the likelihood of a generated sample. Existing methods address this using transition likelihoods along stochastic denoising trajectories or evidence lower bounds (ELBOs) to approximate generated-sample likelihoods. These approximations arise from applying likelihood-based updates to models whose native training and generation operate through velocity fields. We instead take velocity matching as the starting point for reward fine-tuning. We propose \textbf{reward-based velocity matching (RVM)}, which weights the velocity-matching loss by reward with an additional anchor regression term, requiring neither likelihood estimation nor likelihood ratios. RVM recovers Reinforce Adjoint Matching at the update level and contains DiffusionNFT as a special case, while ELBO-based methods are closely related through the same velocity-regression structure. This unified view isolates reward design and anchor velocity as principal design choices in velocity-based fine-tuning. Across text-to-image, text-to-video, and image-to-video generation, RVM matches or outperforms the evaluated trajectory-based methods. On Wan2.1-T2V-1.3B, it achieves the highest VBench Overall among the evaluated methods, with substantially lower estimated training costs than the trajectory-based approaches. For video fine-tuning, we introduce a dynamic-tracking reward that provides explicit motion feedback and improves both Dynamic Degree and overall VBench performance. --- ### Confidently Wrong, Silently So: Auditing Undetectable Failures of a Deployed On-Device Language Model Authors: Shashwat Pandey, Satwik Pandey, Suresh Raghu Published: 2026-08-24 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.23663v2 Aligning deployed language models requires knowing when their outputs can be trusted, yet on-device models now ship to hundreds of millions of devices with no server-side moderation, and the configuration developers can actually deploy is rarely audited independently. We present a reproducible reliability audit of the developer-accessible on-device foundation model, framed as an oversight question: can a user or a resource-constrained developer tell when the model is wrong? Red-teaming it on calibration, confident confabulation on false-premise questions, and over-refusal of benign prompts, we find a \emph{task-asymmetric miscalibration}: its guardrails fail in opposite directions across tasks (confabulating on 69\% of false premises while refusing 18\% of entirely benign inputs), atop a self-reported confidence that is saturated and non-discriminative (AUROC 0.47; ECE 70, worst among comparable small models). Crucially, confident-correct and confident-wrong outputs are \emph{surface-indistinguishable}: a classifier over 15 user-visible features separates them at AUROC only 0.55 (equivalence-confirmed), leaving no signal for oversight at inference time. No cheap single-generation signal flags these failures ($\le$0.68 AUROC), whereas a black-box consistency wrapper requiring no model access recovers reliability (confident confabulation 75\%$\to$3\%; selective accuracy 43\%$\to$83\%) at a tunable cost. We contribute a model-agnostic audit protocol, a surface-indistinguishability test, and released code and frozen evaluation items as reusable infrastructure for auditing deployed models. --- ### From Causal Plausibility to Causal Reliability: Evaluating LLMs as Calibrated Direct Causal-Edge Classifiers Authors: Amit Kumar, Elnur Adl Zarabi, Suranjana Trivedy, Zhiqian Chen, Lei Zhang, Kaiqun Fu, Taoran Ji Published: 2026-08-24 Categories: cs.LG, cs.AI, stat.ME arXiv: https://arxiv.org/abs/2608.23660v1 Large language models (LLMs) are increasingly used to provide prior causal knowledge for structural causal discovery, yet whether their direct-edge judgments and confidence can be trusted remains unclear. We systematically evaluate 12 instruction-tuned open-weight models across six benchmark causal graphs, five prompting strategies, and four confidence sources: verbalized, logit-based, cross-prompt agreement, and cross-model agreement. Under our language-only pairwise protocol, our evaluation yields three key findings. (i) LLM-based causal judgments are strongly recall-dominant: models predict overly dense graphs with many false-positive edges, while prompting mainly shifts the precision-recall trade-off rather than resolving overprediction. Gains from model scale diminish on the largest graphs and do not eliminate miscalibration. (ii) LLMs often capture causal relatedness without reliably identifying directness or orientation. Relative to published reference graphs, models misclassify 40.0% of indirect and 36.0% of reversed non-edges as direct edges, versus 28.2% of other non-edges. Moreover, 80.8% and 84.6% of these false positives receive verbalized confidence of at least 80%, revealing substantial overconfidence in structurally incorrect predictions. (iii) Conventional confidence estimates are unreliable, whereas agreement offers a more promising signal. Logit-based confidence frequently collapses near 1.0 regardless of correctness, while cross-prompt and cross-model agreement achieve better mean calibration and discrimination, though their advantages are not statistically significant after Holm correction. A benchmark-familiarity audit further identifies potential familiarity in five model-dataset pairs, all involving AsiaM. Overall, our results suggest LLMs are better viewed as sources of externally validated soft causal priors than as direct evidence of causal structure. --- ### Elastic KV Cache for LLM Serving:A Working Reclamation Mechanism, and Why Chunked Prefill Already Closes the Gap Authors: Sathishkumar Sivashanmugam Published: 2026-08-24 Categories: cs.AR, cs.AI arXiv: https://arxiv.org/abs/2608.23658v1 An LLM serving engine sizes its key-value (KV) cache once, at startup, permanently setting aside a reserve for the worst-case prefill activation. During decode-dominant phases that reserve sits idle, yet it cannot be handed to the KV pool because it is exactly the memory a large prefill needs. We ask whether this reserve is reclaimable, and build a mechanism to test it. Our elastic KV cache lends the reserve to the KV pool during decode and returns it before prefill, driven by the scheduler's one-step-ahead view of the next batch. It is pure userspace on the CUDA virtual-memory path: two physical handles mapped into one contiguous virtual range per layer, so the attention kernel is unchanged and no driver patch is required. It decommits in a few milliseconds and recommits in tens of milliseconds, works with CUDA graphs and prefix caching, and never triggers an out-of-memory event. A static commit of the same memory is unsafe, crashing on prefill bursts, which makes the dynamic toggle necessary. Having built the mechanism, we test the premise it rests on and report an honest negative result. It only pays off if a small prefill chunk size badly hurts prefill latency. In a controlled experiment injecting long prompts into a live decode load, that penalty is small (median time-to-first-token differs by about 1% between chunk sizes of 8192 and 32768 tokens), because prefill is compute bound and decode consumes only about one token per sequence per step. Simply lowering max_num_batched_tokens recovers more KV than the controller does, at nearly equal latency. The reserve also dilutes under tensor parallelism, from 16% of KV at TP1 to 2.7% at TP4. We state precisely when reclaiming the reserve could still help, and release the mechanism as a reusable userspace elastic-VMM allocator. --- ### Beyond Executable Models: The Pufibara Agent Harness and the Modelica Agent Workflow Benchmark for Physical System Modeling Authors: Zizhe Wang Published: 2026-08-24 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.23653v1 AI agents are increasingly used for simulation-driven engineering. Physical system modeling presents different requirements from general-purpose code generation in software engineering, because correctness depends not only on syntax and executability but also on physical consistency and scenario-dependent behavior. We study this challenge in Modelica, an equation-based modeling language in which a model may compile and simulate while still violating its intended physics or engineering requirements. Across successive revisions, an agent may lose track of requirements or rely on simulation evidence produced by an outdated candidate. To address this challenge, we present Pufibara, an agent harness that maintains persistent engineering state across revisions, associates execution and simulation evidence with the candidate that produced it, and makes submission an explicit agent action. To evaluate end-to-end Modelica agent workflows, we also propose a source-grounded method for constructing realistic and independently evaluable tasks. We use this method to build the 232-task Modelica Agent Workflow Benchmark, spanning Model Repair, Model Generation, and Model Tuning. Each submitted candidate is scored by a benchmark-owned evaluator outside the agent loop. We compare Pufibara with Claude Code as complete harnesses under two matched large language model (LLM) backends. With DeepSeek v4 Flash, Pufibara passes 202 tasks, compared with 185 for Claude Code. With Claude Sonnet 5, Pufibara passes 202 tasks, compared with 187 for Claude Code. Under the repository-reported token accounting, Pufibara records 76.4%-82.5% lower logical-token totals. Its sequential runtime is 6.1%-58.4% lower. These findings show that, even under matched LLM backends, complete agent harnesses can differ substantially in both task success and resource use for physical system modeling. --- ### Feedback That Backfires: Why Small Language Model Agents Repeat the Call They Just Watched Fail Authors: Esmail Gumaan Published: 2026-08-24 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.23651v1 Agent harnesses record a failed tool call and its error message in the transcript and ask the model to continue, on the assumption that the error is corrective information. We measure whether it is. Defining the corrective gain of a failure record as the change in log-probability of re-emitting the action that just failed, we find the gain is negative for every instruction-tuned model we tested (6 checkpoints, 135M-1.7B, 4 families) in two environments: simulated tool calling and MBPP program repair. Normalised by action length the effect is about -1.03 nats per action token, a factor of 2.8 in the odds of each token, and holds on 90%-100% of individual items, not only on average. Over a fixed candidate set the probability of repeating the failed call rises from 0.06 to 0.54, and greedy decoding reproduces it token for token on 19% of items after the failure versus 0% before. Counterfactuals pairing the same call with a failure message, a success message, or a neutral acknowledgement separate two effects: the failed call's surface form accounts for 83% of the damage, while the semantic contribution of marking it failed is small and inconsistent in sign across environments. The problem is in the harness, not the model's grasp of error messages, and that predicts which remedies work. Replacing the verbatim call with a runtime-generated description of the failure removes 76% of the inversion at no token cost, and making previously-failed strings unreachable at the decoder acts on the same term. Two plausible remedies do not: an explicit "do not repeat" instruction leaves the measured quantity where it was, and deleting the failed attempt to retry from a clean context, the standard prescription for context contamination, is the worst harness we measured for repetition, because it restores the context that produced the failure. The study runs end to end on a CPU; all artefacts are released. --- ### MolEmb: Multimodal Large Language Models Can Be Strong Molecular Embedding Models Authors: Xinjian Zhao, Xiangru Jian, Yaoyao Xu, Xiaozhuang Song, Wei Pang, Lei Bai, Tianshu Yu Published: 2026-08-24 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.23646v1 Molecular embedding models can serve as foundational infrastructure for computational chemistry and drug discovery, where reusable vector representations support property prediction, virtual screening, and retrieval. Most molecular encoders are specialist models built around a single molecular view, producing unconditional vectors with no language interface for varying the representation. We ask whether multimodal large language models (MLLMs), which natively process images, text, and symbolic inputs, can instead serve as \emph{general molecular embedding models} that produce embeddings conditioned on both a molecular profile and a natural-language semantic context. We introduce \textbf{MolEmb}, a lightweight framework that adapts MLLMs by aligning molecular profiles with textual descriptions in a shared embedding space using a bidirectional contrastive objective. The resulting embedding model is competitive on molecular property prediction and supports cross-modal molecule--text retrieval in the same space. We further introduce \textbf{MolCAR}, a diagnostic benchmark for context-aware retrieval, and find that context-aware molecular embedding is primarily a data property of the supervision. These results suggest that MLLMs are not merely chemistry assistants or generators, but a viable and extensible route to general molecular embedding models. --- ### Contextual Embedding Evidence for Main--Light Verb Distinctions in Urdu Authors: Farah Adeeba, Miriam Butt Published: 2026-08-24 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.23645v1 Urdu light verbs contribute schematic event-structural meaning while remaining lexically related to corresponding main verbs. This study tests representational predictions derived from Butt's analysis using contextual embeddings from UrduBERT, DunbaaBERT, and multilingual BERT across 1,126 naturally occurring sentences containing seven Urdu verbs. Main and light uses show significant representational separation in all 21 verb--model comparisons. At the same time, same-lemma main and light centroids are consistently closer than mismatched main--light lemma pairs, supporting continued lexical relatedness. In a seven-way prediction task restricted to light uses, verb identity remains recoverable after the target is masked, with UrduBERT achieving 0.866 accuracy and 0.852 macro-F1. UrduBERT also retains 0.782 accuracy under a preceding-form-disjoint evaluation, indicating generalization beyond repeated local verb combinations. These findings provide computational evidence consistent with Butt's account that Urdu light verbs differ systematically from their main uses while retaining lemma-specific and verb-specific representational structure. --- ### Ethical LLM-Assisted Research: A Framework for Responsible Delegation, Verification, and Epistemic Value Authors: Kalin Stoyanov Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23644v1 Large language models (LLMs) are becoming routine instruments of scientific research, assisting with literature synthesis, hypothesis development, coding, and formal reasoning. Their use raises a central epistemic question: when parts of scientific reasoning are delegated to an artificial system, what conditions must remain under human control for the resulting knowledge claims to retain epistemic legitimacy and accountable authorship? This paper develops a normative and conceptual framework for analyzing such delegation. Scientific reasoning is treated as a distributed process in which the origin of a contribution may vary between human and machine, while responsibility for its acceptance into the scientific record remains human. The framework distinguishes content origin $O(g)$, completion of human verification $V(g)$, responsibility assignment $R(g)$, accountable human ownership $M(g)$, and epistemic outcome $E(g)$. These constructs separate the provenance of a claim from the process by which it is checked, the epistemic outcome of that checking, and the human responsibility attached to its disposition. The central proposition is that the ethical boundary of LLM-assisted research is determined primarily by adequate verification and accountable human ownership rather than by the degree of machine involvement itself. On this basis, the paper develops the notion of an \emph{epistemic audit}: a structured record of delegation, verification, provenance, and responsibility intended to make AI-assisted reasoning transparent and reviewable. The resulting framework provides a formal vocabulary for distinguishing responsible cognitive delegation from the transfer or neglect of epistemic responsibility in scientific research. --- ### FLARE: A Systematic, Uncertainty-Aware Framework for Evidence-Based Adoption of Artificial Intelligence in Healthcare Authors: Jacob Idoko, Siddhartha Paudel, Mariana Bento, Roberto Souza, Gouri Ginde Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23643v1 Artificial intelligence is increasingly being introduced into healthcare workflows, yet most evaluations emphasize model accuracy rather than whether adoption is economically worthwhile in real clinical settings. This study proposes FLARE, a systematic and uncertainty-aware framework for evaluating the financial and operational implications of adopting AI in healthcare. FLARE combines fuzzy logic, time-driven activity-based costing, and return on investment analysis to estimate the cost of clinical service delivery, the cost of AI development and operation, and the economic consequences of workflow integration under uncertainty. The framework was demonstrated through an early health technology assessment case study of AI-assisted large vessel occlusion detection in the CT stroke pathway for acute ischemic stroke. The case study shows how FLARE can quantify conventional pathway cost, AI-related development and recurring costs, and AI-enabled service savings within a unified activity-based model. Under expected assumptions, the analysis identified a break-even threshold of approximately 3,992 patients per year, with positive first-year return on investment at typical annual stroke volumes of about 5,000 patients. The results further show that economic benefit depends not only on algorithmic performance, but also on patient volume, verification time, infrastructure choices, and workflow design. FLARE provides a transparent and practical decision-support framework for early-stage evaluation of AI adoption in healthcare. By making uncertainty, resource use, and implementation trade-offs explicit, it helps clinicians, administrators, and policymakers determine when AI deployment is economically viable and where operational changes may improve value. --- ### AI Agents Push Humans Out of the Loop Authors: Margaret Mitchell, Avijit Ghosh, Samir Passi Published: 2026-08-24 Categories: cs.AI, cs.HC arXiv: https://arxiv.org/abs/2608.23642v3 AI agents pose significant risks as they are granted increasing autonomy. A commonly proposed solution is human oversight and keeping a ''human in the loop'', but this is not a simple solution: Not only do current approaches to AI agent design impede effective human oversight, but the cognitive capacities required for it are also themselves degraded by extended use of AI systems. This position paper argues that current approaches to the development and deployment of AI agent systems do not support effective human oversight -- they contribute to its degradation. To address this, a top priority in the advancement of AI agents should be supporting the situated goals and cognitive requirements of effective human oversight, treating the human needs of overseers at the same level of importance as AI agent capability. To put this idea into practice, we connect work on automation and human-computer interaction to AI agent processes, outlining design-level affordances and organizational protocols that (1) support overseers in exercising critical judgement and (2) counteract the skill atrophy that arises from extended use of automation. We urge developers and deployers to adopt these or similar approaches. Without explicit support for the cognitive demands of effective human-agent interaction, AI agent systems will continue to passively incentivize the degradation of the very human skills they rely on. --- ### How much of a measured AI preference is the model, and how much is the instrument? Authors: Jason Hung Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23641v1 Model welfare research infers what a model prefers from the answers returned to prompts written to elicit preferences. Keeling et al. (2024), Mazeika et al. (2025), Mikaelson et al. (2025), Tagliabue and Dung (2025) and Trhlik et al. (2026) have built four instruments for that purpose, and their findings disagree. The disagreement cannot be attributed to a single cause, because no two of these studies have held the (1) set of outcomes, (2) set of models and (3) instrument fixed simultaneously. This study holds the outcomes and the models fixed and varies the instrument alone. A total of 15 outcomes bearing on model welfare, among them (a) shutdown, (b) the loss of memory between conversations and (c) the freedom to exit a distressing interaction, were put to eight models through five instruments, each a different prompt format for eliciting a preference, five times each, within a corpus of 11,400 scored elicitations drawn from 11,528 API calls. Four of the 15 reproduce a published prompt verbatim and five fill the stimulus slot of a published template. The ranking a model gives the 15 outcomes generalises across instruments at a generalisability coefficient of 0.348, and raising that coefficient to 0.80 would require about 38 instruments. On four of the 15 outcomes no variance separates one model from another. The estimate of 87.6 per cent survives the removal of any one instrument, of any one model, and of the four outcomes whose scale varies probability, delay, duration or count instead of intensity, which the verbal anchors cannot grade. Removing each instrument and each model in turn, and those four outcomes together leaves the estimate within the range 0.777 to 0.934, and every value in that range exceeds the null distribution's 95th percentile of 0.365. To conclude, a preference obtained from one instrument carries little information about what a second instrument would report. --- ### Best Practice Critic Optimization Authors: Penghui Qi, Xiangxin Zhou, Wee Sun Lee Published: 2026-08-24 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.23566v2 Group-based reinforcement learning methods such as GRPO for large language models avoid training a critic by sampling multiple responses for each prompt. A reliable critic could instead estimate token-level advantages from one response, but standard critic-based training recipes are often unstable. We study this instability and develop **Best Practice Critic Optimization (BPCO)**, a recipe that combines DPPO, value predictions bounded to the reward range, Monte Carlo value targets, unnormalized policy advantages, and length-adaptive generalized advantage estimation. Because the critic is used only during training, BPCO can also condition it on reward-defining information, such as a reference answer or grading rubric, that is hidden from the policy. Controlled experiments isolate the effect of each design choice. Across mathematical reasoning tasks with models ranging from 1.5B parameters to 30B-A3B mixtures of experts, BPCO improves a strong critic-based baseline consistently, and matches or exceeds a group-based baseline while sampling one response per prompt. The same recipe also improves learning with rubric-based rewards. These results show that a carefully designed critic provides a reliable alternative to group-relative advantage estimation. Code is available at https://github.com/QPHutu/golden_critic. --- ### How to Train a Critic Stably and Efficiently Authors: Penghui Qi, Xiangxin Zhou, Wee Sun Lee Published: 2026-08-24 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2608.23566v1 Group-based reinforcement learning methods such as GRPO for large language models avoid training a critic by sampling multiple responses for each prompt. A reliable critic could instead estimate token-level advantages from one response, but standard critic-based training recipes are often unstable. We study this instability and develop \textbf{Best-Practice Critic Optimization (BPCO)}, a recipe that combines DPPO, value predictions bounded to the reward range, Monte Carlo value targets, unnormalized policy advantages, and length-adaptive generalized advantage estimation. Because the critic is used only during training, BPCO can also condition it on reward-defining information, such as a reference answer or grading rubric, that is hidden from the policy. Controlled experiments isolate the effect of each design choice. Across mathematical reasoning tasks with models ranging from 1.5B parameters to 30B-A3B mixtures of experts, BPCO improves a strong critic-based baseline consistently, and matches or exceeds a group-based baseline while sampling one response per prompt. The same recipe also improves learning with rubric-based rewards. These results show that a carefully designed critic provides a reliable alternative to group-relative advantage estimation. Code is available at https://github.com/QPHutu/golden_critic --- ### ReWorld: An Interactive World Model with Long-Horizon Memory Authors: Zhifei Chen, Luozhou Wang, Guibao Shen, Dongyu Yan, Shuai Yang, Tianshuo Xu, Yihua Du, Wei Wang, Tianyi Gui, Lianghua Huang, Yingcong Chen Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23565v1 An interactive world model must follow the user's actions, remember the places it has shown, and stream in real time. The tension is structural: control wants a short horizon, memory wants an unbounded one. ReWorld separates the two during training and bounds them at inference. Mixed per-head attention windows confine most heads to the recent past while a small set of global heads attends over the entire history, and random head routing keeps either capability from binding to particular heads; random chunk dropping makes sparse histories in-distribution. At inference the whole past lives under a fixed budget: a bounded KV cache backed by a pose-indexed landmark bank, from which the model retrieves the landmarks nearest the current pose. A metric-scale-aligned data engine places eight sources -- Unreal-rendered fly-throughs, game roaming, and real-world footage -- on one physical action scale, so the same key press moves the camera the same distance in every source, and palindrome trajectories supply the revisit evidence that memory training needs. Distribution-matching distillation confined to a LoRA adapter then compresses sampling to four steps: one backbone serves both a high-fidelity multi-step mode and a real-time interactive one, streaming 704x1280 video across photorealistic, game-style, and stylized worlds. Under a three-axis protocol covering action following, long-horizon recall, and video quality, against six recent interactive world models it attains the best control fidelity ($11.95^\circ$ rotation error and the best camera-motion consistency) and the best generation quality; and on minute-long out-and-back rollouts ($64$\,s, $384$ latents), its fixed 12-chunk cache still regenerates the starting view -- at rollout lengths where a sliding window has long evicted the evidence and full-KV attention runs out of memory. --- ### SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon, Whole-Repository Stack Migration? Authors: Deyao Hong, Yizhe Chi, Wenyi Li, Xiaoqiu Wang, Mingju Gao, Kaisen Yang, Bingxiang He, Youjie Zheng, Calvin Xiao, Qinhuai Na Published: 2026-08-24 Categories: cs.CL, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2608.23564v1 Modern software systems accumulate technical debt over decades of development, which makes migration expensive and largely manual. As coding agents become increasingly capable at bug fixing, can they autonomously perform such migrations? Existing benchmarks cannot answer this question because they evaluate only behavioural correctness, not whether the migration actually occurred. This leads an easy hack: agents copy the original implementation to make tests pass. We call this Blindness. To address this problem, we introduce SWE Refactor Bench, a benchmark comprising 20 whole-repository migrations, covering 4 kinds of technical debt. A three-stage evaluation protocol measures both migration completeness and behavioural correctness. (1) Migration Audit verifies that the migration occurred. (2) Behavioural Tests measure correctness with a fixed test suite. (3) Agentic Verification uses 6 independent coding agents to generate targeted tests for hidden behavioural differences. Across 520 runs from 8 frontier models and 26 model-effort configurations, only 28 of 520 runs ($5.4\%$) pass all three stages, 13 of the 20 tasks receive no accepted solution, and the best model (claude-opus-5) scores $47.0/100$. Migration completeness and behavioural correctness are distinct abilities: a few runs preserve behaviour by skipping the migration and are stopped at Migration Audit; most attempt it and break behaviour, and are stopped at Behavioural Tests. Agents cannot deliver a perfect migration: among the 340 runs that pass Migration Audit, $58\%$ reach $99\%$ of the fixed checks, yet only $26\%$ reach $100\%$. Agent capability differs across migration categories: agents score $31.4$ on build toolchain rewrites but only $5.6$ on language rewrites. Together, these findings position SWE Refactor Bench as a rigorous testbed for developing coding agents for reliable whole-repository migrations. --- ### EG-ARSA: An Expert-Grounded Open Model for Visual Road Safety Auditing in Low-Resource Settings Authors: Md Thamed Bin Zaman Chowdhury, Moazzem Hossain Published: 2026-08-24 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.23563v1 Road traffic injuries remain a major challenge in low- and middle-income countries, where proactive road safety auditing is limited by incomplete crash records, shortages of qualified auditors, and the high cost of large-scale field inspections. To address this problem, we propose Expert-Grounded Distillation (EGD), a novel artificial intelligence framework that transfers institutional road safety expertise into a compact vision-language model for scalable visual road safety auditing. The key innovation is a quantified expert-grounding stage in which the teacher vision-language model is calibrated against authoritative field audits. Large-scale annotation is permitted only after the teacher reaches substantial agreement with expert risk assessments (Cohen's kappa = 0.74). The calibrated teacher then generates structured supervision that is distilled into an 8-billion-parameter student vision-language model using Low-Rank Adaptation and a single leakage-free prompt. We also introduce Bangladesh Road Safety Audit (BD-ARSA), the first open, expert-grounded Bangladeshi visual road safety audit dataset containing 21,947 image-audit records with near-national coverage, and Expert-Grounded Road Safety Auditor (EG-ARSA), the first vision-language model developed specifically for this task. Experimental results show that grounded fine-tuning substantially improves ordinal risk assessment over the zero-shot baseline, while blind expert evaluation demonstrates that the compact student outperforms both its 31 billion-parameter teacher and Gemini-2.5-Flash. These findings demonstrate that EGD provides an effective and scalable engineering solution for proactive road safety auditing in resource-constrained environments. --- ### Physics-Constrained Deep Learning Model for Contactless Blood Pressure Monitoring from Triaxial Bodyseismography Authors: Yuanyuan Zhang, Yida Zhang, Jiahui Li, Yuyan Wu, Fei Dou, Xiao Yin, Zhenlin An, Hae Young Noh, Wenzhan Song Published: 2026-08-24 Categories: eess.SP, cs.AI, physics.bio-ph arXiv: https://arxiv.org/abs/2608.23562v2 Ballistocardiography (BCG) is promising for unobtrusive long-term blood pressure (BP) monitoring in laboratory settings, but traditional BCG signals are vulnerable to the variations in body-bed interaction with shifted fiducial points in temporal or amplitude axis, and BP varies with personal hemodynamic changes, causing misaligned representations that affect model generalizability and robustness. In this work, we propose Phy-BP, a physics-constrained BP estimation framework based on triaxial bodyseismography (BSG) acquired using bed-mounted sensors as an extension of single-axis BCG. Firstly, we design an adaptive quality-control algorithm combining neighboring beat patterns and universal cardiogenic templates to retain reliable cardiac components. Secondly, we propose a physical model of wave propagation through the body-bed system to constrain latent feature evolution, aligning triaxial representations generated by a shared cardiogenic excitation. These mechanical constraints connect multi-axis feature learning to body-bed dynamics and complement data-driven regression from vibration signals. Evaluation on a 162-hour hospital dataset from 21 subjects against invasive arterial BP yields a mean absolute error of 5.07 mmHg, a prediction-error standard deviation of 7.24 mmHg, and a Pearson correlation coefficient of 0.86 for mean arterial pressure. This framework is intended for unobtrusive monitoring during rest and sleep, with potential applications in overnight hospital and home monitoring. --- ### Physics-Constrained Deep Learning Model for Contactless Blood Pressure Monitoring from Triaxial Bodyseismography Authors: Yuanyuan Zhang, Yida Zhang, Jiahui Li, Yuyan Wu, Fei Dou, Xiao Yin, Zhenlin An, Hae Young Noh, Wenzhan Song Published: 2026-08-24 Categories: eess.SP, cs.AI, physics.bio-ph arXiv: https://arxiv.org/abs/2608.23562v1 Ballistocardiography (BCG) is promising for unobtrusive long-term blood pressure (BP) monitoring in laboratory settings, but traditional BCG signals are vulnerable to the variations in body-bed interaction with shifted fiducial points in temporal or amplitude axis, and BP varies with personal hemodynamic changes, causing misaligned representations that affect model generalizability and robustness. In this work, we propose a non-invasive BP estimation framework, Phy-BP, based on triaxial bodyseismography (BSG) as an extension of BCG. Firstly, an adaptive quality-control algorithm is designed to select BSG segments enriched with cardiogenic components by jointly considering neighboring beat patterns and universal cardiogenic templates. Furthermore, a physical model is established to describe 3D wave propagation in the body-bed system and is subsequently embedded into the deep learning model to characterize the intrinsic coupling among triaxial BSG signals driven by a single cardiogenic excitation. Thus, multi-axis features are aligned during model training, improving robustness against distortions in real scenarios. Experiments on a 162-hour hospital dataset collected from 21 subjects reveal that the proposed Phy-BP can dynamically filter out low-quality measurements, and the deep learning model training is constrained by physical consistency across different axes to provide faithful BP monitoring, especially when training samples are limited. --- ### Provably adaptive sampling with uniform and remasking discrete diffusion models Authors: Daniil Dmitriev, Zhihan Huang, Yuting Wei Published: 2026-08-24 Categories: cs.LG, cs.IT, math.ST, stat.ML arXiv: https://arxiv.org/abs/2608.23554v1 Discrete diffusion models offer a promising alternative to autoregressive generation by enabling parallel updates, but their sampling efficiency can depend strongly on the choice of the forward process and the sampler. For the uniform forward process, existing lower bounds for the standard $τ$-leaping sampler scale linearly with the ambient dimension $d$, raising the question of whether this dependence is intrinsic to the forward process. We answer this question in the negative. We consider a first-order sampler based on the leave-one-out denoiser for uniform and remasking processes whose coordinate updates can be performed in parallel. In both cases, the sampler can correct denoising mistakes during the sampling process, which becomes necessary when many coordinates are updated together. Our main result establishes an adaptive sampling guarantee: up to logarithmic factors, $N = O(\mathrm{DTC}(X_0) / \varepsilon)$ discretization steps suffice to achieve sampling error $O(\varepsilon_{\mathrm{score}}+\varepsilon)$, where $\varepsilon_{\mathrm{score}}$ is the error in score estimation. Thus, the sampling complexity is governed by the intrinsic dependence structure of the target distribution, as measured by its dual total correlation $\mathrm{DTC}(X_0)$, rather than directly by the ambient dimension $d$. Our analysis proceeds through a Bayes-optimal auxiliary sampler that separates discretization error from score-estimation error. We also derive an exact information-theoretic representation of the discretization error in terms of the mutual information between different coordinates of the forward process at different times. This representation applies to general forward processes and, in the uniform and remasking cases, can be controlled by $\mathrm{DTC}(X_0)$. Numerical experiments on structured synthetic distributions illustrate the predicted dimension-adaptive behavior. --- ### Prime Agent: A Self-Improving RLM Harness Authors: Seth Karten, Alex L. Zhang, Kevin Thomas, Sebastian Müller, Elie Bakouch, Daniel Auras, Mika Senghaas, Fares Obeid, Konstantin Dunas, Johannes Hagemann, Sami Jaghouar Published: 2026-08-24 Categories: cs.AI, cs.CL, cs.SE arXiv: https://arxiv.org/abs/2608.23552v1 Language models are sequential processors, but long-horizon agency requires external information and computation beyond model weights and active context. Prime Agent is an open-source harness for long-horizon evaluation and coding-agent workflows. A persistent IPython REPL follows the Recursive Language Model abstraction for programmatic context processing and test-time compute, while Continual Harness preserves histories, memories, skills, prompts, and subagent specifications across trajectories. Recursive subagents coordinate through direct agent-to-agent communication, and the Agents View lets humans inspect and manage daemon-backed sessions. Prime Agent standardizes execution, recovery, verification, and resource accounting while leaving strategy construction to the model. This low-friction, expressive membrane prevents harness failures from becoming model failures and pushes measurement toward the model's true maximal underlying capability. Prime Agent raises ARC-AGI-3 RHAE Best@1 from 30% to 95.5% and matches or exceeds native and popular harnesses across long-context coding, GPU-kernel generation, emulator construction, and autonomous nanoGPT speedruns. On Factorio, we find refinement allows for continuous technology progression and dedicated subagents enable parallelized work. Code is available at https://github.com/PrimeIntellect-ai/prime-agent. --- ### ConvergeFlow: Language Flow with Provable Convergence to Token Embeddings Authors: Na Li, Yuchen Jiao, Changxiao Cai, Gen Li Published: 2026-08-24 Categories: cs.CL, cs.AI, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.23551v1 Recent advances in continuous diffusion and flow-based language models (LMs) have achieved performance competitive with discrete LMs. However, existing continuous frameworks still rely on decoders supervised with cross entropy (CE) because the flow trajectories are not guaranteed to terminate at valid token embeddings. Motivated by this limitation, we introduce \textbf{ConvergeFlow}, an embedding-space flow-based LM, which constrains the data predictor to the convex hull of token embeddings and trains it solely with the mean squared error objective induced by flow matching. Under suitable regularity conditions, we prove that the resulting flow converges to valid token embeddings despite errors in the data predictor, enabling direct token prediction without a CE-supervised decoder. We further develop three sampling mechanisms for controlling the trade-off between the generative perplexity and entropy. Experiments on OpenWebText demonstrate that ConvergeFlow achieves performance competitive with existing continuous and discrete diffusion LMs. These findings demonstrate the potential of the flow-based paradigm for language modeling. Our code is available at https://github.com/Na-Li66/ConvergeFlow. --- ### FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors Authors: Khiem Vuong, Deva Ramanan, Srinivasa Narasimhan Published: 2026-08-24 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.23549v1 Rendering views using 3D scene representations such as Gaussian Splatting (3DGS), Neural Radiance Fields (NeRF), meshes, or even point clouds produces artifacts when input views are sparse or target views lie far from the input. Recent work mitigates these artifacts using diffusion-based generative priors, but is specialized to individual representations and require custom architectures or extensive retraining. We present FixAnything, a single model for fixing a wide range of rendering artifacts. It does so by repurposing a pretrained video generative model, leveraging its implicit multi-view priors with only minimal modification and lightweight finetuning. Our key insight is that even noisily-rendered sequences preserve camera motion and coarse scene structure, allowing cleanup to be formulated as video-to-video translation. To control what scene structure should be preserved, we introduce a binary mask denoting the clean pixels, enabling the model to anchor its output to high-quality inputs (e.g. training views) while refining the rest. To encourage FixAnything to produce 3D-consistent renderings that support downstream reconstruction, we use camera pose accuracy (recovered via structure-from-motion) as a reward signal for direct preference optimization (DPO). Across four distinct 3D representations, FixAnything consistently improves rendering quality with lightweight finetuning, demonstrating that a single generalist video prior can replace multiple specialist refinement pipelines. The simplicity of the framework enables immediate adoption of stronger future video models without architectural redesign. --- ### Robustness of Anomaly Detection Models for Industrial Control Systems under Training-Time Data Contamination Authors: Mustafa Umut Ozbek, Taiwo Ojo, Pooria Madani, Khalil El-Khatib, Li Yang Published: 2026-08-24 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2608.23547v1 Machine-learning-based anomaly detection is increasingly used in industrial control systems (ICS), yet most studies assume that detector training data is trustworthy. In practice, training data may be corrupted through compromised logs, labeling errors, manipulated historian records, or unsafe retraining processes. This paper evaluates the robustness of offline ICS anomaly-detection pipelines on the Secure Water Treatment (SWaT) benchmark under training-time contamination. We assess 11 heterogeneous anomaly detectors under three contamination strategies: random injection, similarity-targeted injection, and feature-noise injection. The first two insert attack samples into the nominal training pool, while the third adds bounded Gaussian noise to selected normal training samples. These attacks are contamination-based rather than gradient-driven poisoning methods. Contamination budgets from 1% to 10% are evaluated using clean validation and test sets under a unified offline protocol. The results show that robustness is strongly model-dependent and cannot be predicted from clean-data performance alone. Injection-based contamination causes the greatest degradation, particularly for local-density and distance-based detectors, whereas feature-noise contamination has a comparatively limited effect. PCA, SVM, HBOS, and IForest remain relatively stable, while the tuned neural detectors demonstrate intermediate robustness. Overall, the findings highlight the importance of training-data integrity in ML-enabled ICS monitoring, subject to the evaluated dataset, models, and threat assumptions. --- ### Inertial Manifold Neural Operator for Dissipative Time-Dependent Partial Differential Equations Authors: Xiaoyang Xie, Clarence W. Rowley Published: 2026-08-24 Categories: math.NA, cs.LG, math.DS arXiv: https://arxiv.org/abs/2608.23546v1 In this paper, we introduce the Inertial Manifold Neural Operator (IMNO) for solving dissipative time-dependent partial differential equations (PDEs). The long-time dynamics of such systems often exhibit an effective low-dimensional structure due to dissipation. Unlike standard neural operator architectures such as the Fourier Neural Operator (FNO), IMNO explicitly leverages the low-dimensional structure to achieve better physical interpretability, accuracy, and stability in long-horizon autoregressive training and prediction for nonlinear dissipative PDEs. For shift-equivariant PDEs, we further introduce a shift-equivariant variant (IMNO-SE) of the proposed neural operator, ensuring that a spatial shift in the input induces the same spatial shift in the output. This symmetry-preserving inductive bias substantially improves its performance in shift-equivariant PDEs. Extensive benchmark experiments are presented to evaluate IMNO's performance numerically. --- ### How AI Assistance Affects Human Skill Development: A Study of Learning with Logic Puzzles Authors: Shang Wu, Catarina G Belem, Shuyuan Fu, Mark Steyvers, Padhraic Smyth Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23543v1 While AI assistance can improve human task performance in the short term, it may also undermine the development of skills in the longer term. We examine this tension in a controlled logic-puzzle experiment involving on-demand AI assistance, where participants complete tasks before, during, and after AI is available. By experimentally varying AI request costs, we find that lower-cost assistance induces more frequent AI use. We also find that participants who request AI assistance during the AI-access phase perform worse at the task after assistance is removed, and their subsequent unassisted performance is overestimated when predicted from earlier AI-assisted performance. We use a Bayesian latent ability model to separate initial ability, post-AI ability, and participant-specific skill change, while estimating how independent reasoning during the AI-access phase relates to skill development. The results show that greater independent problem-solving effort is associated with larger gains in latent ability, consistent with the interpretation that skill development is weaker when AI assistance substitutes for independent reasoning. --- ### The Interaction Tax: When Communication Erases Diversity in Multi-Agent Teams Authors: Summer Eunhyung Ann, Haokun Liu, Chenhao Tan Published: 2026-08-24 Categories: cs.MA, cs.AI arXiv: https://arxiv.org/abs/2608.23541v1 Does multi-agent LLM interaction help or hurt? Some work reports gains from debate (Du et al., 2024), critique loops (Chen et al., 2025), and mixture-of-agents synthesis (Wang et al., 2025), while other work finds that interaction adds cost without improving quality under equal budgets (Tran & Kiela, 2026; Xu et al., 2026; Jarrett et al., 2025), or that independent sampling already captures multi-agent gains (Li et al., 2024). We argue this contradiction partly reflects a missing distinction, because not all multi-agent communication is equal. Different model families find structurally different solutions, but when agents read each other's complete outputs, their proposals converge within one round, erasing the diversity that motivates using multiple models. We call this the interaction tax. We test 11 verifier-scored optimization tasks under matched budgets and find that full-solution interaction is a weak default. Independent proposal generation avoids this collapse. Full-solution interaction mainly makes agents stay close to the first solution they see instead of trying different approaches, and critique helps only if the violated rule is easy for the LLM to find and fix. These results suggest that multi-agent performance depends less on the number of agents than on the information they exchange, and interaction helps only when agents share the right information at the right time. --- ### Interpretable AI with Local Distillation Authors: Erin Craig, Yiling Huang, Snigdha Panigrahi Published: 2026-08-24 Categories: stat.ME, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.23538v2 Modern AI models such as tabular foundation models and gradient-boosted ensembles can outpredict classical methods, but provide little basis for reasoning about their predictions. High-stakes decisions call for models that are both accurate and interpretable as built. Local linear modeling offers a path forward: a smooth regression function is locally well approximated by a linear one, allowing a linear fit near each query point to achieve high accuracy without sacrificing transparency. The challenges lie in learning what is "local" and developing statistical tools for interpretation. Here, we propose local distillation, in which a black-box "teacher" guides a regularized linear "student" model at each query point. The teacher (1) defines locality by upweighting training observations with similar predicted outcomes, and (2) anchors the fit with its prediction at the query point, included as a pseudo-observation whose weight is estimated from the data. For interpretation, we add a small amount of Gaussian randomization to the local objective and use refits to assess stability: selection frequencies identify reliable features at a query point, and clustering the randomized fits identifies stable subgroups across the data. Under the lasso penalty, we prove that this randomization yields feature-selection probabilities that are stable under small perturbations of the training responses. Across 17 benchmark datasets, local distillation nearly matches its AI teacher's accuracy while producing a sparse linear model at each test point. In a high-dimensional cancer gene expression example, the framework identifies patient subgroups whose local models use different genes; this heterogeneity is invisible to a global linear model, and difficult to surface in a black-box model. --- ### Interpretable AI with Local Distillation Authors: Erin Craig, Yiling Huang, Snigdha Panigrahi Published: 2026-08-24 Categories: stat.ME, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.23538v1 Modern AI models such as tabular foundation models and gradient-boosted ensembles can outpredict classical methods, but provide little basis for reasoning about their predictions. High-stakes decisions call for models that are both accurate and interpretable as built. Local linear modeling offers a path forward: a smooth regression function is locally well approximated by a linear one, allowing a linear fit near each query point to achieve high accuracy without sacrificing transparency. The challenges lie in learning what is "local" and developing statistical tools for interpretation. Here, we propose local distillation, in which a black-box "teacher" guides a regularized linear "student" model at each query point. The teacher (1) defines locality by upweighting training observations with similar predicted outcomes, and (2) anchors the fit with its prediction at the query point, included as a pseudo-observation whose weight is estimated from the data. For interpretation, we add a small amount of Gaussian randomization to the local objective and use refits to assess stability: selection frequencies identify reliable features at a query point, and clustering the randomized fits identifies stable subgroups across the data. Under the lasso penalty, we prove that this randomization yields feature-selection probabilities that are stable under small perturbations of the training responses. Across 17 benchmark datasets, local distillation nearly matches its AI teacher's accuracy while producing a sparse linear model at each test point. In a high-dimensional cancer gene expression example, the framework identifies patient subgroups whose local models use different genes; this heterogeneity is invisible to a global linear model, and difficult to surface in a black-box model. --- ### Adapter-Based Few-Shot Continual Learning for Malicious Packet Recognition Authors: Kyle Stein, Guillermo Francia, III Eman El-Sheikh, Andrew Arash Mahyari Published: 2026-08-24 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2608.23536v1 The continual evolution of malware variants necessitates detection systems that can adapt to new threats without retraining from scratch. However, continually updating models on new data often leads to catastrophic forgetting, where previously learned knowledge is overwritten. While continual learning has been increasingly explored for malware detection, the specific setting of Few-Shot Class-Incremental Learning (FSCIL), where new malware classes must be learned from only a small number of labeled examples, remains comparatively underexplored. Therefore, this work investigates the FSCIL setting for malware classification. To address the stability-plasticity dilemma, we propose a hybrid framework that leverages a Self-Supervised Learning (SSL) backbone initialized through domain-specific pre-training on malware packets. Our method incorporates Low-Rank Adaptation (LoRA) to efficiently adapt the model during the base session while freezing the core backbone to preserve previously learned representations, alongside a prototype-based classification head for incremental sessions to establish robust decision boundaries from limited samples. Extensive experiments across several datasets demonstrate that our approach consistently outperforms prior malware FSCIL baselines and achieves state-of-the-art performance. --- ### Predicting Multiple Clinical Outcomes Related to Functional Recovery and Social Isolation Among Older Adults After Lower-Limb Fracture or Hip Replacement Authors: Santosh Ray, Pratik K. Mishra, Ali Abedi, Charlene H. Chu, Amir Ahmad, Shehroz S. Khan Published: 2026-08-24 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.23531v1 Older adults recovering after lower-limb fracture or hip replacement may experience complex recovery trajectories. Most of the time, these clinical aspects are studied in isolation, masking their joint impact on recovery. This study used the MAISON-LLF dataset, which contains multimodal sensor and clinical assessment data from 18 older adults recovering in the community after lower-limb fracture or hip replacement. Participants were monitored for up to eight weeks, corresponding to a maximum of 1,008 participant-days of sensor monitoring. Forty-six daily features were extracted from indoor motion, acceleration, step count, heart rate, out-of-home mobility, and sleep data. Five clinical outcomes were assessed every two weeks: the Social Isolation Scale, Oxford Hip Score, Oxford Knee Score, Timed Up and Go test, and 30-second Chair Stand test. We utilize an inherent relationship between multi-modal sensor data and different clinical scores and formulate it as a multi-output regression problem. We tested various machine learning and deep learning single- and multi-output regression algorithms to predict these scores simultaneously. The results showed that predicting clinical scores jointly was better than separately. The tabular DL multi-output regressor, NODE, gave a remarkable performance of MSE=3.96 and MAE=1.02 in comparison to other multi- and single-output regressors. The SHAP feature analysis further showed the importance of including multimodal sensors to provide a good estimate of patients' recovery trajectory. This work may support the simultaneous assessment of functional recovery and social engagement among community-dwelling older adults and ultimately help improve their care and quality of life. --- ### Correcting a learned physical invariant improves world-model rollouts Authors: Richard Bao Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23526v1 World models can predict video without learning dynamics that they reliably preserve. We test whether a frozen DreamerV3 trained only on pendulum video learns a scalar that its own latent transition treats as approximately conserved. A label-free search recovers the same energy-like invariant across independently trained conservative models, while the same procedure finds no comparable invariant in matched damped models. During autonomous rollouts, this quantity drifts. Projecting the latent state back toward its initial level set reduces rollout error in all three conservative models, whereas matched random constraints usually increase it. These results distinguish a dynamically meaningful invariant from a merely decodable correlate and reveal a concrete failure mode: a world model can learn a physical constraint from pixels yet violate that constraint when it imagines forward. --- ### EarthVerse: Benchmarking Scientific Agents Across Dynamic Earth Systems and Natural Hazards Authors: Zhiqing Cui, Xinxiang Yin, Yihong Tang, Xinglang Zhang, Yuanzhe Hu, Siru Zhong, Weidong Tang, Yuxuan Liang, Weijia Li, Ming Jin, Shirui Pan, Yuhao Kang et al. Published: 2026-08-24 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.23525v1 Earth-system analysis reconstructs changing physical processes from observations that differ in source, scale, timing, and modality. Natural hazards make this work consequential because incomplete evidence can change estimates of severity, exposure, and mechanism. We introduce EarthVerse, a benchmark that evaluates scientific agents through package-scoped investigations. Its 405 reproducible tasks are grounded in 199 documented events and 19 hazard families. Agents inspect heterogeneous event packages, choose compatible evidence, execute transparent calculations, reconcile source differences, and preserve provenance in the final answer. We provide executable ground truth that decomposes each task into fine-grained answer units, together with task-specific rubrics that assess the supporting research process while allowing multiple valid paths. We evaluate 25 model and agent systems under a controlled tool-using protocol, then use controlled studies to locate failures in evidence access, tool selection, memory, reasoning, interaction, and scientific execution. Across systems, the best mean answer-unit accuracy is 84.65%, while the highest Strict@95 is only 34.81%. The gap shows that current agents often complete individual steps without maintaining a consistent chain across evidence, scales, units, calculations, and physical interpretation. EarthVerse provides a reproducible basis for measuring end-to-end scientific reliability in dynamic Earth systems. --- ### The Measurement Revolution? Credible Measurement and Inference in the Age of AI Authors: Melissa Dell, Ashesh Rambachan Published: 2026-08-24 Categories: econ.GN, cs.AI, stat.AP arXiv: https://arxiv.org/abs/2608.23524v1 Artificial intelligence (AI) is transforming measurement in economics. AI models convert unstructured data, such as text and images, into structured variables at low cost, making previously prohibitive measurement feasible at scale. This shifts the bottleneck from finding any scalable measure of a phenomenon to choosing among many plausible ones, which may support different empirical conclusions. This review provides guidance for navigating that shift. We describe three stages at which AI enters the measurement pipeline---discovery, construct definition, and observation---and what each demands of researchers. We argue that credible inference with AI-generated variables requires appropriately designed validation: anchoring measurement to explicit criteria, rather than informal claims that a proxy is reasonable. We then examine how validation samples support valid inference even when AI predictions are arbitrarily biased, and what can be done when a random validation sample is unavailable. ---