# Research Papers — 2026-08 · bundle page 166 of 240 > Full-text markdown of papers #8251–8300 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/166 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/165.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/167.md) ### UNSPECIFIC: General Constraint Synthesis for Breaking Copy-and-Paste Shortcut in LLM Instruction Following Authors: Jeet Sharma, Balpreet Kaur, Jeremiah Hong, Hamed Zamani, Haw-Shiuan Chang Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09154v1 Large language models (LLMs) are increasingly expected to follow long lists of constraints in complex instructions, and synthesizing instructions from a reference document (i.e., back-translation) is a widely used method to measure/enhance LLMs' ability to follow complex instructions. However, this method introduces a critical loophole: the constraint synthesis model copies text from the reference as a very specific constraint and the evaluated LLM trivially satisfies the constraint by copying its text in the response. To address these issues, we propose UNSPECIFIC, a novel framework that synthesizes constraints common to two similar reference articles to reduce copy-pasting, selectively hardens only trivially satisfied constraints to balance difficulty and naturalness, and evaluates satisfaction on both the generated article and its summary to penalize superficial instruction following. Consequently, we built the UNSPECIFIC benchmark on news, story, and blog domains to analyze the copy-pasting behavior of LLMs. Our results show that our synthesized constraints are not only more challenging (e.g., the satisfaction rate of GPT-5 Mini drops from 90% to 78%) and natural (LLM win-rate gap improves by 30%) from a human perspective but also mitigate the copy-pasting. We also find that a large portion of constraints are satisfied superficially (i.e., not satisfied in the core narrative of the article). The code and datasets are released at https://github.com/JeetDSharma/UNSPECIFIC. --- ### TRACE: TRajectory Attribution for Automated Context Engineering Authors: Yikai Zhao, Pradeep Kumar Misra, Saurabh Pandey Published: 2026-08-10 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.09153v1 Production AI agents fail when their context sources -- system prompts, knowledge bases, tool descriptions, and procedural skills -- contain errors or gaps. Current maintenance relies on manual log review and ad-hoc debugging, creating a scalability bottleneck as interaction volume grows. We present TRACE (TRajectory Attribution for Automated Context Engineering), an automated feedback loop that mines historical agent trajectories to diagnose and remediate context failures. Our key insight is that trajectories are rich with implicit dissatisfaction signals -- user corrections, rephrasing, abandonment cues -- that reveal precisely where context sources failed, without explicit feedback collection. Unlike model fine-tuning, TRACE operates on the context layer, enabling rapid iteration without retraining. We make four contributions: (1) a trajectory mining framework that systematically extracts diagnostic information from historical agent executions; (2) multi-component causal attribution that extends textual gradients from monolithic prompt optimization to heterogeneous context sources (skills, knowledge bases, tools, prompts); (3) exploratory verification, where agents actively read context sources to distinguish content gaps requiring CREATE from stale content requiring UPDATE, achieving 96% operation accuracy; and (4) a reusable simulation methodology and verifiable benchmark addressing the absence of open datasets for context debugging, with a six-category fault taxonomy, ground truth annotations, and a cross-layer verification protocol. On 60 dissatisfaction traces spanning three complexity tiers (up to 16 execution nodes), TRACE achieves 72.7% root cause attribution and 82% end-to-end fix effectiveness, showing that over 80% of context-layer failures can be automatically diagnosed and remediated by mining historical trajectories, an overlooked resource in production systems. --- ### LightAIR: Lightweight Action Inversion and Riemannian Rectification for Text-based Person Anomaly Search Authors: Yulun Zhang, Zixu Li, Zhiwei Chen, Zhiheng Fu, Wenbo Wang, Zihang Qiu, Zhilin Wang, Ruxin Wang, Yupeng Hu Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09152v1 Traditional Text-based Person Search (TPS) is typically limited to matching static appearance attributes, severely neglecting dynamic action information. The Text-based Person Anomaly Search (TPAS) task bridges this gap, requiring models to locate micro-level specific abnormal behaviors while matching macro-level appearance of pedestrians. However, current TPAS methods face fundamental limitations: external explicit pose estimators are fragile in unconstrained surveillance scenarios, and implicit learning encounters visual decoupling failure under pixel-level entanglement, causing dominant appearance information to easily swallow and contaminate subtle action features. Furthermore, performing contrastive optimization on hard negative samples (``same appearance, different actions'') in conventional Euclidean spaces induces severe shortcut learning. To address these, we propose the Lightweight Action Inversion and Riemannian rectification network (LightAIR). First, it introduces textual semantic priors as anchors via a lightweight action inversion operator to extract pure action features, thereby overcoming visual-inherent coupling. Subsequently, it employs orthogonal null-space projection to constrain appearance features within the orthogonal complement space of action features, guaranteeing strict forward decoupling. Finally, we designed a gradient rectification module that computes the Riemannian gradient to constrain the backpropagation trajectory, forcing the gradient flow to update strictly along the tangent space that preserves decoupling properties, thereby cutting off harmful shortcuts. Extensive experiments on the widely used TPAS and TIPR datasets demonstrate that LightAIR significantly outperforms existing state-of-the-art methods. Codes are available at https://github.com/rainy-london/LightAIR --- ### OGG-FR: Orthogonal Gradient Gaming and Frequency Rectification for Unmanned Aerial Vehicle Infrared Image Super-Resolution Authors: Yongsong Huang, Qingzhong Wang, Xiaofeng Liu, Tomo Miyazaki, Yaohou Fan, Shinichiro Omachi Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09150v1 Unmanned aerial vehicle (UAV) infrared image super-resolution aims to recover weak thermal structures for deployment on resource-constrained platforms; lightweight models are therefore preferred, but multi-loss training can be unstable. A common strategy combines pixel-domain and frequency-domain objectives; however, low contrast, limited high-frequency content, and sensor-specific noise often make their gradients weakly aligned or conflicting. To address this optimization ambiguity, we propose Orthogonal Gradient Gaming and Frequency Rectification (OGG-FR), a plug-and-play optimization framework that decomposes the frequency gradient into a redundant parallel component and an orthogonal innovation component relative to the pixel gradient. In the conflict regime, OGG-FR computes a safe base gradient using the Multiple Gradient Descent Algorithm (MGDA) and adds a variance-rectified orthogonal innovation; in the compatible regime, it discards redundant parallel information and injects the orthogonal innovation according to a confidence score estimated from the high-frequency residual. Experimental results on the UAV thermal benchmark show broad gains under BI and BD degradations at $\times 4$ and $\times 8$ scales, while gradient analyses support the effectiveness of the proposed conflict-aware update rule. --- ### RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection Authors: Zhihao Zhang, Gengwei Zhang, Tianlong Chen, Xiaoming Liu Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09147v1 Monocular 3D object detection spans two regimes: closed-set detectors operating within a fixed category vocabulary, and open-vocabulary detectors that localize arbitrary categories by leveraging depth foundation models for 3D geometry. We find that current depth foundation models, despite their strong zero-shot generalization, lack the object-level precision 3D detection demands: substituting a state-of-the-art depth foundation model for a strong detector's predicted depth degrades accuracy, even falling below the detector's own prediction. Rather than pushing detectors or depth models to be more accurate end-to-end, we treat object-level depth refinement as a stand-alone task and present RefineAny3D, a vision-language model that corrects depth without ever predicting a numerical value. Our key insight is that depth error has a direct visual signature in image space: when projected onto the image, a correctly placed box tightly encloses the object, while a too-far box projects too small and a too-close box projects too large. Depth refinement thus reduces to a visual alignment problem rather than a metric regression problem, which we instantiate by extending the VLM's vocabulary with action tokens that replace numerical depth output with categorical decisions, and by supervising the model on a large-scale chain-of-thought dataset that grounds each decision in explicit visual evidence. Applied as a single post-hoc step, RefineAny3D delivers consistent gains across closed-set detectors, open-vocabulary detectors, and 3D auto-labeling tools, and generalizes to novel categories, scenes, and cameras without retraining. --- ### Multi-Submap Implicit Neural SLAM with Local-to-Global Loop Closure for Large-Scale Scene Reconstruction Authors: Tianchen Deng, Chongdi Wang, Nailin Wang, Lei Zhao, Ziqi Ma, Tianjun Zhang, Zhe Liu, Danwei Wang, Hesheng Wang Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09146v1 Neural Radiance Fields (NeRF)-based SLAM has demonstrated impressive results in small-scale scene reconstruction, yet scaling these methods to extensive, complex environments remains challenging due to catastrophic forgetting and accumulated trajectory drift. This paper presents a robust, large-scale neural SLAM system featuring a multi-submap architecture and a dual-tier loop closure mechanism. Specifically, we propose a progressive mapping strategy that dynamically allocates neural submaps to maintain high-fidelity representations without memory explosion. For robust pose estimation, an optical-flow-based tracking module is integrated to handle aggressive motions. To address global consistency, we introduce a local-to-global loop closure framework leveraging the foundation model for high-performance global descriptor extraction, significantly enhancing relocalization accuracy under varying viewpoints. Furthermore, an inter-submap online distillation algorithm is designed during back-end optimization to enforce geometric and appearance consistency across overlapping submap boundaries. To validate the system, we developed a customized handheld mechatronic platform and conducted extensive evaluations on both public benchmarks and our large-scale indoor-outdoor datasets. Experimental results, including direct deployment on an onboard computing unit, demonstrate that our approach outperforms state-of-the-art neural SLAM methods in reconstruction quality and localization robustness, providing a scalable solution for real-world robotic perception and digital twinning. We will release the code publicly on \href{https://github.com/dtc111111/MSN-SLAM}{https://github.com/dtc111111/MSN-SLAM} . --- ### Right Answer, Wrong Heat: Explanation-Aware Evaluation and Thermal-Grounded Feedback for MLLMs on Infrared Images Authors: Yongsong Huang, Xiaofeng Liu, Tomo Miyazaki, Yaohou Fan, Shinichiro Omachi Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09145v1 General-purpose multimodal large language models (MLLMs) are increasingly applied to infrared images, where they are commonly scored by answer accuracy alone. However, a correct answer does not ensure that the model's explanation is grounded in infrared thermal evidence. We introduce an explanation-aware evaluation framework that separates answer correctness, output-level explanation groundedness, and thermal grounding for infrared visual questions. Using a Dual-LLM Consensus Judge with a preliminary human-anchor calibration check, we find that correct answers can still rely on weak or visible-light evidence; withholding the original infrared image and showing only a visible-like rendering erodes thermal grounding with little accuracy change; and this erosion is observed most strongly for more capable models but disappears when infrared remains available. We further propose Thermal-Grounded Feedback (TGF), a training-free feedback loop that diagnoses explanation-side failures and revises the explanation while preserving the selected answer. On local paired-input validation, TGF improves explanation-side grounding without changing answers. These findings suggest that future trustworthy MLLMs for infrared scene understanding should be evaluated and developed to produce thermally grounded explanations rather than merely accurate answers. --- ### UniMoFlow: Grounding Instruction-Driven 3D Human Motion Editing in Generation Authors: Yilei Hua, Beibei Jing, Ce Zheng, Hanyu Zhou, Yawei Luo, Wei Yang Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09143v1 Instruction-driven editing of 3D human motion requires precise spatiotemporal localization, rich semantic grounding, and strict preservation of unmodified content. Existing methods either resort to training-free adaptation of generative models or rely solely on triplet supervision; however, adaptation often yields suboptimal control, and manually curated triplet datasets remain severely limited in scale and semantic diversity. To overcome this bottleneck, we ground motion editing directly within text-to-motion generation across data, architecture, and inference. At the data level, we develop a closed-loop synthesis-and-verification pipeline that produces Omni-MoEdit, a large-scale dataset spanning body-part, amplitude, temporal, action, and style edits. At the architectural level, we introduce UniMoFlow, a unified latent flow-matching model that shares broad semantic and kinematic knowledge between generation and editing. At the inference level, SAFE (Source-Anchored Flow Editing) complements UniMoFlow with controllable, source-anchored refinement. Furthermore, we augment standard evaluations with semantics-aware metrics to account for valid edits that inherently deviate from a single ground-truth reference. Extensive experiments demonstrate improved target-text alignment, edit effectiveness, and cycle consistency, while maintaining competitive source fidelity and text-to-motion generation quality. --- ### An Agentic Generative Large Language Model for Treatment Planning of Colorectal Cancer Authors: Mengxian Lyu, Cheng Peng, Tim Jang, Ang Li, Mengyuan Zhang, Ziyi Chen, Leighton Elliott, Tianshi Liu, Lidice Galindo, Chiranjeevi Sainatham, Oscar F. Borja-Montes, Kaleb E. Smith et al. Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09142v1 Treatment planning in precision oncology requires synthesizing heterogeneous patient information with rapidly evolving clinical guidelines to ensure guideline-concordant care. While large language models (LLMs) show promise in many diagnostic tasks, their adoption for high-stakes treatment planning is hindered by complex reasoning, adherence to timely clinical guidelines, and safety concerns. In this study, we present GatorOnco, an agentic LLM for colorectal cancer (CRC) treatment planning. GatorOnco is developed using a total of 282 billion tokens of biomedical text, including healthcare system-scale clinical text comprising 166 billion tokens from UF Health. We implemented a domain-adaptation method that integrates pre-training, model merging, a two-stage post-training approach, and agent-based reinforcement learning. An agentic retrieval-augmented generation (RAG) approach dynamically integrates time-sensitive clinical guidelines into the reasoning process. In a blind, randomized clinical evaluation conducted by five UF Health oncologists, GatorOnco significantly outperformed open-source LLMs (P < 0.01) and achieved expert-level performance comparable to UF Health oncologists. Compared with expert oncologists, GatorOnco received significantly higher ratings for readability (4.46 vs. 4.19, P < 0.01) and completeness (3.91 vs. 3.52, P < 0.01), while showing statistically comparable performance in correctness (4.09 vs. 4.11, P = 0.921), currency (4.04 vs. 3.98, P = 0.478), and safety (4.22 vs. 4.22, P = 0.999). These findings demonstrate that integrating agentic reasoning with large-scale domain adaptation can help bridge the gap for generative AI in high-stakes cancer treatment planning. --- ### Beyond Direct Identifiers: Probabilistic Privacy Risk Estimation for Privacy-Conscious LLM Query Delegation Authors: Li Siyan, Zhou Yu, Julia Hirschberg Published: 2026-08-10 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2608.09140v1 Recent work on protecting privacy during user-LLM interactions often focuses on direct, explicit identifiers: the personally-identifiable information (PII) captured by standard detectors. One such approach is Privacy-Conscious Delegation (PCD), where a local LLM acts as an intermediary. However, privacy risk does not stem solely from explicit identifiers but also PII-free self-disclosures, leaving users identifiable through combinations of quasi-identifying traits. We investigate a probabilistic variant of PCD, where we augment its objectives with an LLM-driven probabilistic estimation of k-anonymity. To facilitate this, we first create the PUPA-SD dataset, which contains naturalistic user queries with self-disclosure. Our preliminary results indicate that optimizing PAPILLON on PUPA-SD improves quality on unseen conversations across a variety of local models and produces the best privacy-utility balance for Llama-3.2-3B, while smaller models struggle to jointly optimize quality and privacy. We propose k-anonymity as a useful auxiliary metric for tackling PCD. --- ### CodecArena: Codec Quality Assessment via Visual Reinforcement Learning Authors: Jiaye Fu, Weiqi Li, Qiankun Gao, Yanchen Zhao, Xiandong Meng, Jian Zhang, Siwei Ma, Jiaqi Zhang Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09139v1 Video coding is advancing into the low and ultra-low bitrate regime, driven by end-to-end codecs that replace the hand-crafted pipeline with jointly optimized neural networks and generative codecs that exploit the priors of video generation models. Yet the dominant metrics, LPIPS and DISTS, measure feature and texture similarity rather than content fidelity: a reconstruction that hallucinates a wrong face or blurs text into convincing strokes can still score well, even when a human rejects it instantly. To address this, we propose CodecArena, the first vision-language framework for video coding quality assessment, casting codec evaluation as source-conditioned comparative reasoning between a reference and its reconstructions. We optimize CodecArena with Facet-GRPO, a visual reinforcement learning scheme that aligns pairwise codec preferences while grounding the verdict in five fidelity facets: identity, objects, text, texture, and temporal consistency. Its facet-anchored reward uses automatically derived facet directions as weak anchors, rather than human per-facet labels, to prevent any single sub-score from dominating the holistic preference and to yield interpretable fine-grained quality judgments. To support training and evaluation in this underexplored regime, we construct two complementary resources: CodecArena-1K, a fully automatic preference dataset of 1,500 comparison groups built from traditional, neural, and generative codec reconstructions with fused vision-language and objective supervision; and CodecArena-Bench, a human-ranked benchmark with source-disjoint videos for fair out-of-domain evaluation. Extensive experiments demonstrate that CodecArena achieves state-of-the-art agreement with human judgments on source-disjoint content across diverse codecs and bitrates, surpassing perceptual metrics and prior vision-language evaluators. --- ### SpeedTuning: Speeding Up Policy Execution with Lightweight Reinforcement Learning Authors: David D. Yuan, Tony Z. Zhao, Kaylee Burns, Chelsea Finn Published: 2026-08-10 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2608.09138v2 While learned robotic policies hold promise for advancing generalizable manipulation, their practical deployment is often hindered by suboptimal execution speeds. Imitation learning policies are inherently limited by hardware constraints and the speed of the operator during data collection. In addition, there are no established methods for accelerating policies learned via imitation, and the empirical relationship between execution speed and task success remains underexplored. To address these issues, we introduce SpeedTuning, a reinforcement learning framework specifically designed to enhance the speed of manipulation policies. SpeedTuning learns to predict the optimal execution speed for actions, thereby complementing a base policy without necessitating additional data collection. We provide empirical evidence that SpeedTuning achieves substantial improvements in execution speed, exceeding 2.4x speed-up, while preserving an adequate success rate compared to both the original task policy and straightforward speed-up methods such as linear interpolation at a fixed speed. We evaluate our approach across a diverse set of dynamic and precise tasks, including pouring, throwing, and picking, demonstrating its effectiveness and robustness in enhancing real-world robotic manipulation. Videos and code are available at https://daivdyuan.github.io/speed-tuning/ --- ### Bright-Channel Retinex Enhancement with a Conditional Overdispered-Noise Analysis Authors: Jongpil Jeong Published: 2026-08-10 Categories: cs.CV, eess.IV arXiv: https://arxiv.org/abs/2608.09137v1 I present a training-free low-light enhancement method that combines local bright-channel illumination estimation, Retinex division, and edge-preserving denoising. For a fixed illumination estimate, a conditional Negative -Binominal psueduo-count method characterises the heteroscedastic noise amplified by division. The unconstrained reflectance ratio is the pixelwise maximum-likelihood estimate, with a boundary solution for zero-valued observations; the implemented estimate additionally applies illumination filtering and range clipping. The NB model is a diagnostic noise analysis rather than a calibrated sensor model, and the final fixed-bandwidth bilateral filter is an empirical approximation rather than the exact Bayesian solution. On the LOL-v1 dataset, the methodobtains mean PSNR/SSIM of 17.74dB/0.739, the highest values among the evaluated with conventional methods. A 400X600 image is processed at approximately 43 FPS on an Apple M2 Pro CPU. --- ### When Latents Forget Pixels: Restoring Fidelity in Diffusion Transformer Super-Resolution Authors: Yu Shi, Yuyao Zhang, Yu-wing Tai Published: 2026-08-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.09133v1 Image super-resolution (SR) with large generative models has recently achieved remarkable perceptual quality, yet maintaining fidelity to the LR observation remains challenging. In particular, we observe that diffusion transformers (DiTs) built on latent representations suffer from a critical limitation: the compression bottleneck of the VAE weakens fine-grained spatial information, leading to hallucinated details that are weakly grounded in the input image. In this work, we revisit generative SR from a representation perspective and propose a pixel-grounded super-resolution (PGSR) framework that preserves LR-observed pixel evidence before VAE compression and reuses it throughout restoration. Instead of relying solely on the compressed latent condition, PGSR extracts pre-VAE pixel evidence from the upsampled LR image and reuses it at two stages. First, Condition-Side Trajectory Guidance fuses LR-derived pixel evidence with the latent LR condition to guide the latent restoration trajectory. Second, Decoder-Side Pixel Grounding injects multi-scale pixel features into the frozen VAE decoder to ground the final rendering with LR-observed cues. To efficiently adapt large pretrained DiT models, we keep the latent autoencoder and main flow-matching backbone frozen, and train only lightweight restoration modules. We further study an efficient local-window attention variant for improved high-resolution efficiency and scalability. Extensive experiments demonstrate that PGSR improves the realism--fidelity trade-off and produces more faithful, visually convincing results than existing latent generative SR approaches. --- ### MARA: Flow-Matching-Guided Multi-Agent Resource Allocation for Computational Resource Efficient Learning Authors: Hanye Zhao, Muning Wen, Yong Yu, Weinan Zhang Published: 2026-08-10 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.09130v1 Allocating limited computation among concurrent learning tasks is difficult when each task must reach a target loss before a deadline but its required training effort is unknown. Existing approaches combine online loss prediction with adaptive resource allocation, yet commonly treat computation as continuously divisible throughput. We instead study a practical setting in which tasks arrive over time and computation is provided by discrete nodes. This setting introduces both uncertain demand and constrained sequential decisions. We propose MARA, which predicts future loss trajectories with conditional flow matching and coordinates compute nodes through a cooperative multi-agent autoregressive policy. A potential-based progress reward supplies intermediate training feedback while preserving the undiscounted task-completion objective. Across in-distribution, reinforcement-learning, and vision workloads, flow matching reduces remaining-resource prediction error relative to weighted least squares. At the scheduler's training load, MARA completes 63.46% of tasks on average, 8.54 percentage points above strong baseline Learning with Adaptive Resource Allocation (LARA), and remains ahead under unseen heavier workloads. --- ### Social Gym and SPaRTan: Benchmarking and Improving LLM Social Reasoning via Multi-Agent Game Tournaments Authors: Keyu He, Xuhui Zhou, Maarten Sap Published: 2026-08-10 Categories: cs.CL, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2608.09128v1 LLM agents are increasingly deployed in multi-agent social settings where they must cooperate, negotiate, and adapt to other agents. Measuring and improving these social skills is hard because, unlike math or logic, social interaction offers no objective ground truth: evaluations fall back on LLM judges, which are costly, subjective, and noisy, and models get no reliable signal to learn from. To address both, we first introduce Social Gym, an environment of 21 multi-agent social games (e.g., Werewolves, Resistance, Spyfall) whose rule-decided outcomes make agent performance verifiable and objective, with an Elo tournament that produces a cross-game leaderboard. Benchmarking experiments show that while GPT-5-mini tops the leaderboard, no model excels at all games uniformly or in all game roles, pointing to limitations of social reasoning. Motivated by this, we additionally propose SPaRTan (Self-Play and Reflect-Transfer), a training-free self-improvement loop: a model plays a game, reflects on its trajectories and their outcomes to produce a transferable playbook, and applies that playbook in subsequent games. Our results show that SPaRTan playbooks help GPT-5-mini agents level their performance on weaker roles, but largely do not improve Qwen3-32B's performance. Together, Social Gym and SPaRTan offer a reproducible, verifiable foundation for measuring and improving LLM social reasoning without weight updates. --- ### Subjective Multi-Bias Detection with Large Language Models Authors: Ruiyu Li, Zhiying Zhu Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09126v1 In this project, we delved into the pervasive challenge of bias detection within the text content. More specifically, our focus lies on the identification of subjective bias, a type of bias that introduces improper attitudes or portrays a statement at odds with the actual truth. The subjective bias can jeopardize the authenticity and reliability of texts, leading to misconceptions and potential social tensions, especially when expressed through offensive language. Following prior work (1), we tackled with three different types of subjective biases in text: (1) framing bias with the use of one-sided words or phrases containing a particular point of view; (2) epistemological bias which includes subtle linguistic features that can affect the believability of the texts; (3) demographic bias with word/phrase usage under presuppositions of a particular demographic factor (i.e., gender or religion). In terms of the data we utilize, the input consists of texts that may harbor subjective biases. The output is a classification or annotation that reveals the presence or absence of such biases within the provided content. More specifically, we detected three different types of multi-span biases in corpus WIKIBIAS (2) with more than 4,000 sentence pairs from Wikipedia edits. The data is labelled by bias type for span pairs with the following categories: (1) framing bias, (2) epistemological bias, (3) demographic bias, and (4) no bias. The project codes are released at https://github.com/HoningJade/LLM-Bias-Type-Classification. --- ### ChronoState: Hidden Elapsed-Time Conditioning for Temporal-State Action Selection in Frozen-Backbone Language Models Authors: Sam Siavoshian, Omar Ramadan, Amir K. Saeed, Benjamin A. Johnson, Amin Mohamed El-Amin Diab, Benjamin M. Rodriguez Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09124v1 Temporal decisions in language-model systems often depend on both symbolic task state and elapsed wall-clock time, such as cache expiration, job completion, quota resets, deadlines, or stale sessions. We study whether elapsed time can be supplied as a non-token, system-side scalar and composed with visible symbolic state by a frozen-backbone language model. We introduce ChronoState, a compositional temporal-state benchmark in which symbolic state appears in the prompt, elapsed seconds tau are supplied through a hidden chronometric-injection channel, and the model selects a forced-choice temporal action. Here, "hidden" means hidden from the user-visible token sequence, not from model computation. Using Qwen2.5-3B-Instruct as a frozen bf16 backbone with a 31-dimensional sinusoidal-plus-log time encoding, gated FiLM residual modulation, and a rank-8 LoRA action surface, hidden-time CI reaches 0.9305 +/- 0.0134 accuracy and 0.9410 +/- 0.0103 balanced accuracy. No-time and shuffled-time controls fall to 0.5511 +/- 0.0042 and 0.3323 +/- 0.0097, respectively, with high shuffled-time wrong-state consistency supporting causal dependence on the injected scalar within the trained distribution. Generalization remains strong for held-out templates, durations, and multi-constraint compositions, but held-out quota-family transfer is weak at 0.5065 +/- 0.0559, while a fair prompt+LoRA timestamp baseline reaches 0.9893 +/- 0.0052. Thus, ChronoState supports a narrow conclusion: hidden elapsed time can be composed with symbolic task state under direct supervision, but does not establish autonomous time tracking, broad unseen-family abstraction, or superiority over prompt-injected timestamps. --- ### RISE-RL: Rubric-Informed Selective Exploration for Open-Ended Reinforcement Learning Authors: Jinkun Hou, Zhuo Liu, Huimin Ren, Hongsheng Xin, Pan Zhou, Kun Zhan Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09123v1 Aligning Large Language Models (LLMs) for open-ended tasks is challenging because responses must satisfy multidimensional criteria without following a single correct generation trajectory. Existing rubric-based reinforcement learning (RL) methods compress fine-grained criterion-level feedback into scalar rewards, making persistent capability gaps difficult to target under limited on-policy exploration. We propose $\textbf{RISE-RL}$ (Rubric-Informed Selective Exploration), which uses repeatedly missed rubric criteria to elicit privileged trajectories that are difficult to discover through unguided exploration alone. RISE-RL retains only trajectories whose complete-rubric reward exceeds the mean reward of natural rollouts, and then re-evaluates them under the original prompt to emphasize behaviors that remain weakly supported by the natural policy. The resulting guidance signal is optimized through a separate auxiliary objective and removed once its additional benefit diminishes. Experiments with 4B and 14B models across writing, chat, health, and science show that RISE-RL achieves the highest mean score on every evaluated benchmark under guidance-free evaluation. Compared with standard Rubric-RL, it improves the average score by 1.3 points at the 4B scale and $\textbf{3.3 points at the 14B scale}$, including a $\textbf{6.0-point}$ gain on CreativeWriting-V3. It also improves creative-writing diversity and yields gains on objectively scored medical and scientific benchmarks. These results indicate that selective internalization through reward filtering and policy support shaping is effective for open-ended reinforcement learning. --- ### Visual Distortion Detection in UGC Images Using Large Multimodal Models Authors: Ziheng Jia, Yingji Liang, Jiaying Qian, Xiongkuo Min Published: 2026-08-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.09122v1 The localized depiction of perceptual quality has long been a crucial, yet underexplored, challenge in image quality assessment (IQA). Existing approaches based on large multimodal models (LMMs) predominantly rely on text-driven supervised fine-tuning (SFT). However, this training paradigm exhibits notable limitations in detection accuracy. Moreover, synthetically distorted images, which are often used as the primary training data source, show a significant generalization gap when deployed in real-world scenarios; thus, the \textbf{synthetic-to-authentic (\textit{S2A})} problem represents a critical challenge. Motivated by these issues, we propose \textbf{\textit{VIGIL}}, which leverages the LMM architecture for precise visual distortion detection. From a candidate pool of over 1000K samples, we construct the \textbf{\textit{VIGIL-140K}} training set, which consists of over 140K distorted images. These images are obtained through rigorous quality filtering and carefully crafted distortion injection, covering 8 major synthetic distortion categories. Our model leverages different layers of the large language model (LLM) decoder, treating them as \textit{multiple detectors} that perform synchronous distortion detection using multi-level features. Additionally, we retain distortion cues from predictions assigned to the non-distortion class, which helps mitigate the ambiguous foreground-background (\textit{FG-BG}) separation commonly encountered in the \textit{S2A} problem. After post-processing, our model consistently outperforms strong baselines on both in-domain synthetic distortion detection and \textit{S2A} tasks. --- ### MELLON - Multimodal Enhanced LLM for Online Navigation Authors: Ruiyu Li, Haoyang Cai, Zhitong Guo, Tong Hu Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09121v1 Web navigation agents are capable of addressing various types of tasks on different websites. Current baselines on web navigation are either unimodal or lack strong reasoning abilities given multimodal inputs. Focusing on the WebShop benchmark, a real-world website simulation, we explore the alignment of text and images, as well as multimodal reasoning and planning abilities, to enhance the performance of web navigation agents. We propose three innovative multimodal enhancements: Multimodal Enhanced LLM for Online Navigation (MELLON), VQAgent, and Multimodal Ranker. MELLON demonstrates a significant improvement in task completion accuracy, with a 9.26% increase after just one epoch of training. Our findings suggest the necessity of further exploration into multimodal approaches, with a focus on more extensive training and alignment strategies to enhance the effectiveness of web navigation agents. --- ### Motif 3: Technical Report Authors: Junghwan Lim, Joon Son Chung, Sungmin Lee, Wai Ting Cheung, Gihun Cho, Minsu Ha, Sangho Kang, Beomgyu Kim, Dongseok Kim, Jangwoong Kim, Taehyun Kim, Taewhan Kim et al. Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09119v1 We introduce Motif 3, a decoder-only Mixture-of-Experts language model with 314 billion total parameters and 13.2 billion activated per token. Each sparse MoE layer contains 384 routed experts, with eight selected per token. This fine-grained sparsity provides substantial expert capacity while limiting computation. Motif 3 is built around Grouped Differential Latent Attention (GDLA), which integrates grouped differential attention with the compressed key-value representation of Multi-head Latent Attention. The architecture further incorporates modified manifold-constrained hyper-connections, Expert Specific PolyNorm activations, and multi-token prediction to improve optimization stability, expert specialization, and inference efficiency. We pretrain Motif 3 on approximately 12.5 trillion tokens spanning web documents, STEM, code, mathematics, multilingual content, and domain-specialized corpora. Expert-balancing and numerical-stabilization techniques support stable training at scale, while selective MXFP8 computation and communication, memory-efficient fused kernels, and window-aware context parallelism enable training with context lengths up to 256K tokens. Our post-training pipeline combines general supervised fine-tuning, six specialist teachers trained with reinforcement learning, a software-engineering teacher trained with supervised fine-tuning, and Multi-teacher On-Policy Distillation. The resulting unified model consolidates complementary capabilities in reasoning, coding, tool use, professional work, long-context understanding, calibrated abstention, and instruction following. Across a broad evaluation suite, Motif 3 demonstrates competitive performance against leading open weight models, including strong results on long-horizon agentic tasks, mathematical reasoning, scientific knowledge, and hallucination-sensitive evaluation. --- ### A Probabilistic Circuit-Induced Pseudo-Metric for Out-of-Distribution Detection Authors: Bhumika K, Vidhya S, Narayanan C Krishnan Published: 2026-08-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.09117v2 Probabilistic Circuits (PCs) are tractable generative models whose internal nodes encode a hierarchy of probabilistic summaries over different variable scopes. Existing PC-based out-of-distribution (OOD) detection methods ignore this hierarchy, reducing the entire circuit to the scalar likelihood (or its uncertainty) computed at the root. We introduce Hierarchical Likelihood Vector (HLV), a representation whose entries are the likelihoods associated with selected PC nodes and define the Hierarchical Likelihood Distance (HLD), a PC-induced pseudo-metric that compares the probability distributions through the expectations of their HLVs. We show that HLD is an integral probability metric over a function class naturally induced by the PC and develop a principled goodness-of-fit hypothesis test for unsupervised OOD detection. Unlike existing approaches, the trained PC alone serves as the representation of the in-distribution: no held-out in-distribution data are required at deployment. We further show that the quantities required by the hypothesis test can be computed exactly, directly from the trained circuit, yielding an approximate analytic decision threshold. Experiments on tabular and MNIST datasets demonstrate that exploiting the hierarchical probabilistic summaries encoded through the PC improve OOD detection over root-likelihood, uncertainty-, typicality- and kernel-based baselines, while naturally localizing distribution shifts to the PC nodes responsible for the shift. --- ### RAVEN-Eval: Rubric-Guided Automatic Evaluation for AI Video Generation Models Based on LMM Preference Judgement Authors: Ziheng Jia, Jiaying Qian, Zicheng Zhang, Xiaorong Zhu, Lancheng Gao, Xiongkuo Min Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09111v1 AI video generation has advanced rapidly and entered widespread commercial use. As a result, quality differences among videos produced by state-of-the-art AI video generation models~(AIVGMs) have become increasingly difficult to discern using conventional evaluation criteria, such as visual fidelity and semantic instruction following. Meanwhile, human evaluation now requires more expertise and sustained attention, substantially increasing annotation costs. This calls for automated evaluation that can reliably distinguish fine-grained differences among advanced AIVGMs with minimal human intervention. To address this challenge, we present RAVEN-Eval, a rubric-guided automated evaluation framework for AIVGMs, built primarily on the LMM-as-a-judge paradigm. Through an automatic task curation and quality-filtering pipeline, RAVEN-Eval curates 150 text-to-video~(T2V) tasks and 100 image-to-video~(I2V) tasks, and systematically collects more than 4,500 AIGVs. At its core, RAVEN-Eval adopts rubric-guided automated LMM preference judgement, in which LMM judges conduct pairwise comparisons according to task-specific rubrics. It further introduces an anchor-based model insertion approach to reduce the evaluation cost of incorporating new models. Finally, we evaluate 20 high-performance AIVGMs, as well as the judging capabilities of 13 LMM judges, and establish the RAVEN-Eval Leaderboards. Overall, RAVEN-Eval paves a scalable path for automatic and trustworthy evaluation of rapidly evolving AIVGMs. --- ### View-Adaptive Renderer for View-Consistent 2D-to-3D Generation Authors: U-Chae Jun, Jaeeun Ko, Jiwoo Kang Published: 2026-08-10 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2608.09110v1 Reconstructing 3D shapes from a single image remains a fundamental yet challenging problem in computer vision. Traditional monocular 3D generation pipelines typically synthesize multiple views from a single input image before applying Neural Radiance Field (NeRF)-based reconstruction. However, inherent projective ambiguities often produce visual discontinuities across generated viewpoints, leading to inaccuracies in reconstructed 3D models. Current solutions either incur significant additional computational burdens or fail to adequately resolve practical inconsistencies between synthesized views. To address these limitations, we propose a novel viewpoint-adaptive neural rendering framework that enables robust 3D reconstruction even when given partially inconsistent multi-view inputs. Our approach introduces view-adaptive neural renderers that independently correct viewpoint-dependent errors while simultaneously sharing a global feature backbone to preserve structural coherence. Furthermore, we propose a self-attention fusion module that adaptively integrates multi-view information, ensuring geometric consistency without relying heavily on indirect regularizations or computationally intensive methods. Through extensive experiments, we demonstrate that our method consistently improves 3D reconstruction fidelity. Importantly, our approach achieves near state-of-the-art performance without diffusion-based SDS supervision, relying primarily on photometric rendering loss with lightweight attention regularizers. This balance between accuracy and efficiency makes the proposed framework highly practical for real-world applications. --- ### Different Feedback, Different Updates: Selective Self-Learning from User Interactions for Large Language Models Authors: Xuanchen Li, Haitao Li, Yujia Zhou, Qingyi Pan, Heng Wang, Yiqun Liu, Min Zhang, Qingyao Ai Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09109v1 User feedback offers natural supervision for persistent LLM improvement, but a single message may support multiple behavioral changes with different scopes of generalization. We introduce SLIFT, a selective self-learning framework built on a task-relative view of user feedback. SLIFT decomposes each feedback message into atomic components and interprets each component relative to the original task as Fix, Spec, or Null: requirements for task validity, compatible condition-specific refinements, or content with no reliable positive update direction. To incorporate each change at the appropriate scope, SLIFT trains two complementary LoRA adapters on a shared frozen backbone: a Generalist that consolidates Fix requirements into default behavior through feedback-conditioned self-distillation, and a Specialist that observes only the task and Generalist response to supply residual guidance for applicable, unmet Spec refinements. Null components induce no positive update. Across backbones, SLIFT achieves strong performance on both MemoryBench and WildFB, with targeted analyses further examining its underlying mechanisms. We release our code at https://anonymous.4open.science/r/SLIFT. --- ### LexKairos: Benchmarking Legal Temporal Capabilities in LLMs Authors: Chenyang Li, Zejia Feng, Yuqin Huang, Yuxiao Ye, Huiyuan Xie Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09106v2 Large language models (LLMs) have demonstrated strong performance across a wide range of legal tasks. In legal practice, time is a critical concept that governs the validity of statutes, the progression of legal cases, and the enforcement of procedural deadlines. However, legal temporal capabilities remain underexplored in existing legal AI benchmarks. To address this gap, we propose LexKairos, a comprehensive benchmark for evaluating the temporal capabilities of LLMs in the Chinese legal context across three dimensions: statutory temporal knowledge, case temporal modeling, and statute-case temporal reasoning. LexKairos comprises nine sub-tasks drawn from real-world Chinese judicial cases and statutes. We conduct systematic evaluations of eight LLMs under multiple inference settings, including vanilla, Chain-of-Thought (CoT), and thinking modes. Our results show that Gemini-3-Flash achieves the strongest overall performance, yet even the best-performing model exhibits notable limitations on tasks demanding precise time-sensitive statutory metadata recall or complex reasoning in time limits, indicating that legal temporal knowledge and reasoning remain open challenges for current LLMs. Data and code are available at https://github.com/thunlp/LexKairos. --- ### Multitask Scanning Probe Microscopy Authors: Aditya Raghavan, Yu Liu, Ian Mercer, JP Maria, Sergei Kalinin Published: 2026-08-10 Categories: cond-mat.mtrl-sci, cs.LG, physics.ins-det arXiv: https://arxiv.org/abs/2608.09104v1 Scanning probe microscopy provides nanoscale access to structural, electrical, electromechanical, magnetic, and mechanical properties of materials. Its increasing use for wafer-scale characterization and combinatorial materials exploration creates a need to distribute measurements efficiently across large spatial domains. This is particularly important when available modalities differ in acquisition time and potential for tip and sample damage, making exhaustive multimodal mapping over spatial grids impractical. Here, we demonstrate multitask scanning probe microscopy, a live, closed-loop workflow in which a multitask Gaussian process learns spatial and cross-modal relationships and autonomously selects both the next measurement location and the next experimental protocol. The approach is implemented on an automated large-sample atomic force microscope and demonstrated on a composition-spread AlScN wafer using tapping-mode and Dual AC Resonance Tracking (DART) measurements. Paired initial measurements establish the relation between the tasks, after which noncoincident measurements are used to update both response landscapes. The resulting workflow extends active learning in scanning probe microscopy from spatial sampling to autonomous allocation of measurement modalities and provides a basis for combining rapid, weakly perturbative imaging with slower contact, electrical, electromechanical, magnetic, or spectroscopic measurements. --- ### Contrastive Mask Fidelity: Reference-Free Auditing of Ground-Truth Masks in Remote Sensing Semantic Segmentation Authors: Shuaishuai Cao, Shuwei Peng, Meng Tang, Min Huang, Youjin Wang, Jie Chen, Jing Ouyang, Zhiwei Zhai Published: 2026-08-10 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.09101v1 Semantic segmentation models are trained and evaluated against human-drawn masks, yet remote-sensing annotations are often coarse, incomplete, or misaligned; high overlap scores may then reflect agreement with imperfect labels rather than faithfulness to the image, creating an evaluation paradox. We introduce Contrastive Mask Fidelity (CMF), a training-free, reference-free metric that scores competing class masks directly against image evidence. CMF composites keep and erase counterfactual views of each mask and asks a frozen vision-language judge whether class evidence is concentrated inside the mask and absent outside. We validate CMF on controlled mask corruptions, then audit 10,731 image-class pairs across ten remote-sensing benchmarks using candidate masks from Seg-Probe, a training-free open-vocabulary probe built on SegEarth-OV3 that outperforms prior baselines on nine of ten datasets. The audit reveals systematic, class-dependent annotation distortion: man-made classes such as buildings, roads, and cars favor the candidate mask on 62-85% of pairs, whereas ambiguous land cover more often favors human annotations. On a blinded three-annotator consensus, CMF matches expert judgment on 81% of pairs, exceeding keep-only scoring, model confidence, and a trained label-quality baseline. Finally, conservative class-wise arbitration yields supervision that improves cross-domain transfer over raw annotations and matched replacement controls, positioning CMF as a scalable tool for auditing ground truth rather than presuming it infallible. --- ### Real Data Closes Synthetic-to-Real Gap in Optical Chemical Structure Recognition Authors: Yani Guan, Dengpan Dong, Zi Wei, Shuang Luo, Dan Hannah, Yumin Zhang, Kang Xu Published: 2026-08-10 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2608.09100v1 Millions of chemical structures appear in patents and papers only as drawings, and using that information at scale requires reading the drawings. OCSR appears nearly solved on synthetic images yet remains difficult on real documents: the starting recognizer, Qwen2.5-VL-7B, exceeds 91% accuracy on synthetic renders but falls below 16% on three real-world benchmarks (ACS, CLEF-IP, USPTO). To identify the main source of improvement, 21 recognizers were fine-tuned on mixtures of synthetically rendered structures and labeled real depictions from patents, journal figures, and hand-drawn collections, varying the vision language model (VLM) base, the fraction of real training data, and the vision-tower adaptation strategy. Labeled real training images make the largest difference. For Qwen2.5-VL, ACS exact match rises from 0.15 with no real data to 0.37 at 9.5% and 0.46 at 50.2%; a controlled experiment across three base models reproduces the trend. A vision-tower LoRA, in contrast, does nothing for Qwen (+0.00, paired p=1.00), substantially helps InternVL3-8B (+22.8 to +34.6 pt), and modestly helps GLM-4.1V-9B (+1.0 to +9.6 pt), so its value depends on the base model. The best configuration reaches 0.96 exact match on clean renders and 0.49, 0.65, 0.84, and 0.76 on ACS, CLEF-IP, UOB, and USPTO, respectively. Gaps between base models are largest without real data (0.21), shrink to 0.06 at 70% real data, and reorder the ranking; base model and real-data mixture must therefore be selected together. Small-scale experiments on handwritten image-to-LaTeX recognition and chart-to-table conversion show that base-model rankings also vary beyond chemistry. More generally, model and adaptation choices for visual structure recognition should be evaluated on the target task. --- ### RAVEN: Frozen Random Graph Reservoirs with Physics-Informed Interaction Fingerprints for Protein-Ligand Binding Affinity Prediction Authors: Qingyang Zou, Jiaye Huang, Hangbo Xie, Jiayue Yin, Youyi Song, Jinfeng Liu Published: 2026-08-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.09099v1 Quantitative estimation of protein-ligand binding affinity from three-dimensional complex structures is a fundamental task in structure-based computational chemistry and molecular modeling. Reliable prediction remains challenging because available structure-affinity data are limited, experimentally heterogeneous, conformation-dependent, and sensitive to dataset partitioning. RAVEN (Randomized Atomistic Views with Ensemble Neural Reservoirs) utilizes a multihead reservoir of independently initialized and fully frozen atomistic graph encoders to generate diverse structural projections without end-to-end optimization of the graph representation. These projections are integrated with a deterministic physicochemical interaction fingerprint and processed by heterogeneous supervised readers, including neural and tree-based regressors, whose outputs are combined through validation-based nonnegative fusion. The random reservoir expands structural feature coverage across independent encoder realizations, whereas the explicit physicochemical descriptors and heterogeneous readers contribute complementary information and distinct inductive biases. Evaluation on a similarity-isolated PDBbind 2020R1 split reconstructed using GEMS similarity resources, together with the protected CASF-2016 subset, demonstrated strong predictive performance. The results indicate that frozen multi-view graph representations, explicit physicochemical statistics, and heterogeneous model fusion provide a robust and flexible framework for protein-ligand binding-affinity prediction. --- ### SI-Edit: Toward Sketch-Instruction Guided Local Image Editing with Pixel-Level Precision Authors: Weixin Ye, Wei Wang, Hongguang Zhu, Xuecheng Nie Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09097v1 Despite rapid advances in generative models, achieving pixel-level precision in sketch-based image editing remains a persistent challenge, particularly for fine-grained local deformations. This gap stems primarily from the critical shortage of high-quality, publicly available benchmark datasets that jointly provide geometric constraints and semantic instructions. To address this issue, we first introduce **SI-Data**, a high-quality dataset specifically designed for instruction-guided local sketch editing. We develop an automated pipeline leveraging Multimodal Large Language Models (MLLMs) to synthesize comprehensive quadruplets comprising original images, local geometric sketches, semantic instructions, and corresponding edited images. By providing both reliable spatial anchors and explicit semantic intent, SI-Data uniquely enables collaborative spatial-semantic learning. Building upon this, we propose a collaborative framework called **SI-Edit** that integrates semantic instructions with precise geometric constraints. Furthermore, to address the lack of standardized evaluation, we establish a comprehensive set of metrics designed to measure both structural fidelity (e.g., sketch-to-edge alignment) and semantic adherence. Experimental results demonstrate that SI-Edit provides more reliable structural control than baselines for sketch-based image editing, and achieves precise, pixel-level local refinements aligned with user intent. The data and code are released on the (project page)(https://github.com/ywxsuperstar/SIEdit). --- ### Evo-Bench: Can Language Models Improve Agent Harness? Authors: Lisheng Huang, Chen Yang, Hao Zhou, Huatong Song, Zongchao Chen, Ran Le, Yang Song, Wayne Xin Zhao, Tao Zhang Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09096v2 Large Language Models (LLMs) have driven rapid progress in autonomous agents, yet standard evaluations remain confined to static task solving. An emerging frontier is harness evolution---the agent's capacity to autonomously optimize its own operating harness. However, systematically benchmarking this capability remains challenging, as existing evaluations fail to isolate harness improvements from base model strength, prevent task-specific overfitting, or capture long-horizon iterative research. To address these challenges, we introduce Evo-Bench, the first benchmark designed to evaluate models' intrinsic harness-evolving capabilities across Search, Office, and General agent domains. To rigorously isolate this capability, Evo-Bench employs a novel harness-guided construction framework: it leverages auxiliary-task evolution to identify tasks genuinely sensitive to framework improvements, followed by sensitivity-aware stratified splitting to ensure robust cross-suite generalization. Extensive evaluations across nine frontier and open-weight models reveal that top models achieve massive absolute gains reaching 16.6 points, closely approaching state-of-the-art human-engineered baselines. Crucially, while autonomous evolution outpeforms artificial harness in General tasks and excels in Search tasks, it struggles in Office tasks that demand highly specific processing workflows. Furthermore, our analysis exposes critical temporal anomalies like early saturation, while demonstrating that the synthesized harnesses act as highly transferable reasoning structures, consistently boosting diverse policy models. --- ### Who Bridges Safety? Identifying and Targeting Cross-Lingual Shared Safety Pathways Authors: Shuyi Miao, Wangjie Qiu, Pengyang Shao, Canran Xiao, Fei Shen, Zhiming Zheng, Tat-Seng Chua Published: 2026-08-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.09095v1 Uncovering the internal mechanisms underlying the safety capabilities of large language models (LLMs) is crucial for developing trustworthy artificial intelligence. Currently, mechanistic interpretability studies on multilingual safety are largely confined to local components, such as isolated neurons. However, this static and fragmented perspective overlooks the synergy among components and fails to elucidate how safety signals dynamically propagate within the model to drive safety decisions ultimately. In this work, we move beyond isolated neurons to identify and target the cross-layer functional pathways formed during safety signal propagation, thereby uncovering the mechanisms driving the cross-lingual safety gap. Specifically, we first identify monolingual safety pathways and validate their impact on refusing harmful requests. Subsequent cross-lingual analyses reveal a sparse subset of cross-lingual shared safety pathways, confirming that this intersection acts as the internal bridge transferring safety capabilities from high-resource (HR) languages to non-high-resource (NHR) languages. Building on these mechanistic findings, we propose a pathways-targeted alignment method based on the cross-lingual shared safety pathways. Experimental results show that updating only a small fraction of pathway parameters significantly improves safety in NHR languages while largely preserving the model's general capabilities. --- ### The Announcement Carries the Cue: Markup, Boundaries, and the Notation of Pre-Training Corpora Authors: E. M. Freeburg Published: 2026-08-10 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.09093v1 How a document's arrangement is written down, its notation, is a training variable that no dataset card records. The field has established that text-extraction choices change model behaviour, and has never once measured the notation of what those choices put into the corpus. We define clean-window survival, a deterministic count of how much of a stream still demands the boundary inference, and measure notation on three fronts. What corpora carry: a census of thirteen public corpora, where survival falls to 0.153 in a vision-converted PDF slice against 0.889 in C4; the scarce resource is not unmarked text but long unmarked text; a pre-registered supply test finds what remains institutional, not consumer. Our own pre-registered prediction failed: converters do not fabricate structure on prose, and that null forced the reliability mechanism that survives it. What readers use: across five base models spanning 0.6B to 8.2B and two pipelines, deleting a structural announcement makes the following prose measurably harder to predict, while swapping its notation moves nothing. That zero does not make notation unimportant; it relocates the variable: the operative cue is the announcement, not the sigil. What writers impose: a bounded null. Base models do not impose the marked register above the authored baseline, and handed prose with every announcement deleted they do not put one back, at a rate indistinguishable from zero against an authored reference of zero. We ship the format those measurements imply: the pure frame, paragraphs in authored order, every announcement deleted into a reversible sidecar, mixed against the marked copy over announcement presence rather than notation. Choose format operators by the capability they train, not by the fidelity they preserve, and record extractor identity and survival on data cards. --- ### TLDChoiceNet: Quantitatively Choosing a Transfer Learning Dataset Authors: Jing Ning, James D. Braza Published: 2026-08-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.09091v1 Transfer learning is particularly useful in settings with limited training data, and within image classification it is common to transfer learn upon massive datasets like ImageNet , CIFAR-100, or COCO . Qualitatively, it seems a transfer learning dataset should have both more classes and more examples per class than the fine tuning dataset; however, a quantitative method to choose the best transfer learning dataset does not currently exist. In this paper, we design TLDChoiceNet, a model to choose the best transfer learning dataset given a fine tuning dataset by predicting the test-set accuracy after fine-tuning. A simple version 1 achieves 0.154 MSE on the test dataset, while a version 2 leveraging an ImageNet pre-trained ResNet50 v2 embedding with per-class information attains a 5X lower MSE of 0.031. We further design two metrics that enable an unsupervised method of choosing an optimal transfer learning dataset: distribution distance (DD), which linearly regresses against fine-tune accuracy with an R2 of 0.89, and average class correlation (ACC), which improves the R2 to 0.97. Our results underscore that a dataset's low-level statistics can explain the transfer learning effect, and that using a pre-trained ImageNet can embed different classes further apart in latent feature space. --- ### A Multi-Scale Temporal Framework with Dynamic Fusion for EEG-Based Emotion Recognition Authors: Stefanos Gkikas, Yang Guo, Guangliang Li, Raul Fernandez Rojas, Giorgos Giannakakis, Randy Gomez Published: 2026-08-10 Categories: cs.AI, cs.LG, cs.SD arXiv: https://arxiv.org/abs/2608.09088v1 Mixed emotions represent a clinically relevant but still underexplored target for automatic emotion recognition. EEG provides millisecond-level access to neural activity, yet most EEG pipelines analyze the signal through a single temporal window, thereby fixing the temporal structure available to the model. This study introduces a multi-scale temporal framework for EEG-based emotion recognition. The EEG waveform is decomposed into windows of one or several durations, processed by a shared attention-based encoder, and integrated through a dynamic fusion module that assigns sample-specific weights across temporal scales. The framework is evaluated under a subject-independent protocol in binary and three-class settings, with the three-class task including the mixed affective category. The best results are 65.22% for the two-class task and 45.43% for the three-class task. Both are obtained with three-scale dynamic-fusion configurations and remain substantially above the full-signal baseline. The best-performing temporal scales differ between the two tasks. Dynamic fusion outperforms concatenation in the highest-scoring two-class configuration and slightly exceeds it in the highest-scoring three-class configuration, although these multi-scale settings require substantially more computation than the full-signal baseline. --- ### Learning human joint torques from pixels Authors: Chen Chen, Rui Cheng Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09083v1 Estimating human joint torques from visual observations is a key step toward bringing biomechanical analysis from controlled laboratories to real-world movement scenarios. Existing torque estimation methods typically depend on surface electromyography, motion-capture markers, force plates, or simulated imitation data, which limits their applicability to ordinary RGB images. In this work, we introduce VID, a vision-based inverse dynamics dataset and benchmark for predicting human joint torques directly from real monocular images. VID contains 63,369 synchronized frames with real human images, kinematic annotations, anthropometric attributes, and OpenSim-derived dynamic labels, providing paired visual and biomechanical supervision for real-image inverse dynamics. We further define a standardized evaluation protocol covering overall torque estimation, joint-specific analysis, and action-specific prediction. To establish a strong reference model, we propose VID-Network, which combines pose-pretrained spatial probabilistic features, marker regression, and temporal torque inference to recover joint torques from image sequences. Experiments on VID show that VID-Network achieves an overall mPJE of 1.7612 N$\cdot$m/kg, improving over the best compared baseline by 39.81\%, and obtains the lowest error across all evaluated joint types and most action categories. VID establishes a first practical benchmark for vision-driven human inverse dynamics and provides a foundation for studying biomechanical inference in less constrained environments. --- ### F2STNet: Fair and Federated Spectral-Temporal Modeling for Graph Forecasting Authors: Jiayi Zhang, Jinfeng Xu, Hewei Wang, Siyuan Cen, Haidong Huang, Yiyao Zhan, Zheyu Chen, Jinjiang You, Ai Jian, Edith C. H. Ngai Published: 2026-08-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.09082v1 Spatiotemporal prediction on graph-structured data is central to traffic forecasting and environmental monitoring, yet decentralized and heterogeneous data complicate both sequence modeling and collaborative training. We propose F$^2$STNet, a federated forecasting framework that combines truncated graph-Fourier features, a lightweight diagonal state-space temporal encoder, graph convolution, and Fairness-aware Federated Aggregation (FFA). The spectral branch exposes graph-frequency structure, while the state-space layer models long temporal dependencies with linear complexity in the sequence length. FFA adjusts the FedAvg prior using client validation losses and an increasing fairness schedule. Experiments on PeMS04, HZMetro, and KnowAir show favorable forecasting accuracy relative to the evaluated baselines; federated experiments on PeMS04 additionally improve worst-client and client-dispersion metrics. --- ### When Confidence Fails: Overconfidence in LLMs under Uncertainty and Missing Clinical Information Authors: Maryam Tahermazandarani, Adnan Mahmood, Fahmida Islam, Quan Z. Sheng Published: 2026-08-10 Categories: cs.CL, cs.AI, cs.HC, cs.LG arXiv: https://arxiv.org/abs/2608.09080v1 Large Language Models (LLMs) have achieved strong performance in medical question answering and clinical reasoning tasks. However, their reliability under uncertainty remains poorly understood which raises critical concerns for deployment in high-stakes clinical settings. In such environments, incorrect predictions are inherently risky, but confident incorrect predictions can be particularly harmful as they may mislead clinical decision-making. In this paper, we conduct a systematic behavioral analysis of LLMs under clinical information uncertainty. We propose an evaluation framework based on the MedMCQA dataset consisting of two complementary uncertainty settings. First, we introduce linguistic uncertainty cues through prompt modifications to simulate ambiguous clinical contexts. Second, we construct an answer removal setting, wherein the correct option is deliberately excluded mandating the model to recognize insufficient information and abstain. We analyze both model accuracy and confidence behavior using multiple calibration metrics including calibration gap, Expected Calibration Error (ECE), and Unsafe Confident Error Rate (UCER) across 500 medical questions. Our results reveal a consistent failure mode, i.e., although accuracy degrades under increasing uncertainty, model confidence remains misaligned with accuracy. This leads to a substantial increase in unsafe confident errors, indicating that model confidence remains largely insensitive to clinically meaningful information loss. Furthermore, we observe significant variation across models in their ability to abstain when the correct answer is unavailable, with some models persistently producing high confidence hallucinated answers. These findings expose critical limitations in the epistemic reliability of current LLMs and highlight the need for uncertainty aware evaluation methods prior to their deployment in clinical workflows. --- ### Personalized Federated Learning via Variance-Aware Nonparametric Empirical Bayes Authors: Jae Ho Chang, Arnab Auddy, Subhadeep Paul Published: 2026-08-10 Categories: stat.ML, cs.LG, stat.ME arXiv: https://arxiv.org/abs/2608.09074v1 We develop a new approach to Personalized Federated Learning across heterogeneous clients using Nonparametric Empirical Bayes (NPEB). Leveraging the asymptotic normality of local parameter estimates obtained from Empirical Risk Minimization or M-estimation, our method formulates these estimates as noisy observations to estimate an unknown shared prior via Nonparametric Maximum Likelihood. A key challenge in applying NPEB in this setting is that existing approaches assume known fixed variances, which is not true in practice. To address this, we introduce a Variance-Aware Nonparametric Empirical Bayes (VANEB) framework that leverages the parameter-dependent asymptotic variance of local M-estimators. A key technical contribution is a generalized Tweedie's formula for this heteroskedastic setting. We then establish non-asymptotic error rates for density estimation in the average squared Hellinger distance and derive an oracle denoising inequality that provides error bounds for our estimator. While our theoretical guarantees are rooted in the asymptotic regime of M-estimators, we empirically explore heuristic extensions of VANEB to modern federated learning settings involving Deep Neural Networks (DNNs). For DNNs, we propose VANEB-head and VANEB-FT, which personalize the last fully connected layer via an NPEB step using an approximate diagonal variance estimator. We show that our method has strong performance on popular vision datasets MNIST and CIFAR-10, using a convolutional neural network architecture. --- ### A Unified Issue Resolution Benchmark for Requirement Clarification, Planning, and Code Generation for Coding Agents Authors: Xin Zhou, Chun Yong Chong, Kisub Kim, Yun Peng, Rui Shu, Zihan Wu, Xu Han, Guowen Yuan, Zeyang Zhuang, Jounghoon Kim, Jeongjin Ju, Seongmin Ju et al. Published: 2026-08-10 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.09072v1 Large language model-powered coding agents are increasingly used to modify existing code repositories, for example, by adding features or fixing bugs. Yet existing repository-level benchmarks typically evaluate only whether the final patch passes tests. Satisfying a user request requires a long chain of interdependent reasoning and decisions: an agent must recover explicit and implicit requirements, formulate a repository-grounded implementation plan, and translate it into correct code. A pass/fail outcome cannot characterize how an unsuccessful trajectory diverges from the requirements and implementation process needed for a correct patch. To address this gap, we introduce SWE-RPG, a repository-level benchmark that combines executable patch evaluation with validated ground-truth references (GTs) for (1) Requirement Clarification and (2) Implementation Planning. These intermediate GTs support retrospective, GT-aligned diagnosis of complete coding-agent trajectories across clarification, planning, code generation, and artifact submission. SWE-RPG comprises 163 tasks from 31 Python and Java repositories, including 113 bug fixes and 50 feature additions. We evaluate 3 coding agents, including Claude Code, Codex, and OpenCode, with 6 large language model backends, including Claude-Sonnet-5 and GPT-5.6-Terra. Results show that the evaluated popular coding agents still struggle to implement user requests in existing repositories, achieving an average resolved rate of only 31.5% on SWE-RPG. Intermediate-GT diagnosis further identifies implicit requirement recovery as the main bottleneck, accounting for 24.5%--46.0% of agent runs. This result suggests implicit-requirement recovery as a key candidate direction for improving coding agents. The benchmark data and evaluation code are available at https://github.com/Xin-Zhou-smu/SWE-RPG-Bench. --- ### Two-Step MV-DeepONet: Probabilistic Operator Learning for Uncertainty Propagation Driven by Random Input Fields Authors: Yupei Nie, Lei Wang, Jiasen Liu Published: 2026-08-10 Categories: math.NA, cs.AI arXiv: https://arxiv.org/abs/2608.09071v1 Forward uncertainty propagation in complex physical systems can induce structured covariance across field-valued outputs. For a probabilistic surrogate, the total predictive covariance comprises the covariance of conditional means across input realizations and the average conditional predictive covariance. Probabilistic DeepONet (Prob-DeepONet) provides lightweight uncertainty quantification by predicting pointwise Gaussian means and variances in a single forward pass, but its conditional predictive covariance is restricted to a diagonal form. To represent cross-location conditional dependence without explicitly parameterizing a full high-dimensional covariance matrix, we develop a two-step mean-variance DeepONet (two-step MV-DeepONet) through two principal modifications. First, two-step training is used to decouple output-basis learning from the input-to-coefficient mapping, together with basis orthogonalization and subspace rotation. Second, Gaussian probabilistic modeling is transferred from the high-dimensional physical output space to the low-dimensional rotated coefficient space. Mapping these probabilistic coefficients through the shared basis induces a generally non-diagonal conditional predictive covariance in the physical output space while retaining single-pass inference. A Frobenius-norm error decomposition and corresponding upper bound identify low-rank covariance compressibility, trunk-subspace approximation, finite-sample statistical error, and coefficient-space covariance estimation as the principal factors governing covariance recovery. Numerical experiments on three representative problems governed by partial differential equations (PDEs) and a hypersonic blunt-body aerothermal problem show improved generalization, more structured uncertainty bands, and accurate recovery of off-diagonal correlation patterns compared with Prob-DeepONet. --- ### Model the Edit, Not the Image: Visual Autoregressive Editing from a Source-Centric Perspective Authors: Hongyi Fang, Chuwen Xie, Benjia Zhou, Yu-Xuan Qiu, Chenggong Hu, Zhibin Wang, Chao Chen, Jianbin Qin, Rui Mao Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09057v2 Next-scale visual autoregressive models (VARs) have emerged as a powerful generative paradigm, producing high-quality images through efficient coarse-to-fine prediction. However, their potential for text-guided image editing remains largely underexplored. Existing training-free VAR editing approaches often formulate editing as target-conditioned regeneration guided or constrained by the source image, and may rely on inversion, test-time optimization, attention control, or user-provided masks. This generation-centric formulation does not fully exploit the multiscale source representations provided by VARs and may introduce additional computation or intervention. We instead take a source-centric perspective on VAR editing, in which the encoded source image tokens serve as the primary visual state and the editing process focuses on condition-induced changes. Based on this perspective, we propose \textbf{EditMod}, which compares source- and target-conditioned predictions under a shared autoregressive context, treats their difference as a scale-wise editing direction, and applies it as a residual update to source tokens at selected scales. Experiments show that EditMod achieves leading source-image fidelity while maintaining strong text alignment, and completes end-to-end editing of a 1K image in only 1.57 seconds on a single A100 GPU without per-image preparation. --- ### Diagnosing as Cardiologists Do: ECG Agents with Doctor-Grounded Priors for Clinical Reasoning Across Diseases and Populations Authors: Hongxiang Gao, He-yang Xu, Yuwen Li, Minghui Zhao, Zhipeng Cai, Xingyao Wang, Chenxi Yang, Jianqing Li, Chengyu Liu Published: 2026-08-10 Categories: eess.SP, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2608.09053v1 Cardiologists interpret electrocardiograms by localizing waveform components, measuring rhythm and interval patterns, and translating these structured observations into diagnostic evidence. Whether this expert reading process can serve as an effective prior for ECG agents remains unclear. To address this question, we introduce LuminaECG, a clinically structured ECG reasoning framework that reformulates ECG interpretation as measurement-grounded visual reading. ECG signals are rendered on standard electrocardiographic grid paper to preserve the spatial and scale cues used in clinical reading. P-wave, QRS-complex, and T-wave boundaries are explicitly delineated, and color-coded segmentation decomposes the waveform into discrete visual measurement primitives. A general 2B vision-language backbone is then trained with low-rank supervised fine-tuning to associate these primitives with diagnostic reasoning, without architectural modification. Across open, proprietary, and ECG-specialist zero-shot baselines, LuminaECG improves both waveform measurement and diagnostic recovery. It reaches a clinically meaningful reader tier on the CODE-test benchmark, transfers across geographically diverse ECG datasets without retraining, and generates reports whose structure contains an emergent prognostic signal. These findings suggest that effective ECG agents require not only larger models, but supervision that preserves the alignment between measurable waveform evidence and clinical knowledge. --- ### Triple Expert Learning from Noisy Labels for Semi-Supervised Vision Foundation Model Adaptation Authors: Xuanyu Liu, Zheng Fang, Hongyang He, Yundi Hong, Daizong Liu Published: 2026-08-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.09052v1 Semi-supervised adaptation of vision foundation models (VFMs) commonly freezes the pretrained backbone and updates lightweight modules such as LoRA. However, pseudo-labels have mixed reliability, and a single LoRA adapter must absorb reliable, ambiguous, and noisy gradients in the same low-rank space. This can make VFM adaptation sensitive to pseudo-label noise. We propose \textbf{TriNoL}, a \textbf{Tri}ple-expert learning framework from \textbf{No}isy \textbf{L}abels for semi-supervised VFM adaptation. TriNoL routes unlabeled samples into three confidence regions and assigns them to three LoRA experts: a Positive Expert for high-confidence pseudo-labels, an Alignment Expert for medium-confidence ambiguous samples, and a Negative Expert for low-confidence noisy samples. The VFM backbone remains frozen, and only the LoRA experts and classifier head are updated. By separating different pseudo-label reliability regions into specialized adaptation paths, TriNoL improves robustness to noisy supervision while keeping the training cost low. --- ### Security and Privacy Taxonomy Generation from Mobile App Reviews Authors: Moghis Fereidouni, Vinaik Chhetri, Umar Farooq, A. B. Siddique Published: 2026-08-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.09049v1 Mobile app reviews are a rich, continuously renewing source of how users experience privacy and security, yet existing taxonomies of these concerns are hand-crafted and cannot keep pace with the evolving nature of the data. Automating taxonomy construction is the natural response, but scalability is the core challenge: current LLM- and clustering-based methods are developed for scientific corpora of a few thousand documents and do not extend to app review collections numbering in the hundreds of thousands. We address this gap in two ways. First, we filter app reviews for privacy- and security-related content, yielding a comprehensive corpus of over 600K reviews. Second, we introduce TaxoScale, a pipeline that handles taxonomy construction at this scale by extending an expert-defined taxonomy via Recursive Hierarchical Clustering and LLM-based node naming. TaxoScale outperforms strong automatic-taxonomy baselines on path, level, coverage, and novelty metrics, and discovers novel branches absent from prior taxonomies. --- ### GeoAI-based post-segmentation quality validation of building footprints via spatial feature engineering Authors: Shah Imran Ahsan Chowdhury, Kazi Jihadur Rashid, Rajsree Das Tuli, Rahul Saha, Bulbul Ahammad Published: 2026-08-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.09048v1 Deep learning-based building footprint extraction from high-resolution imagery often produces topologically inconsistent vectors unfit for direct GIS database ingestion. To address this, we present a multidomain GeoAI quality control framework that automates error detection to systematically purify vector footprint databases. Candidate footprints were generated across five UAV survey sites in Bangladesh using U-Net (ResNet-34) and SAM-LoRA (ViT-B). The extracted raster masks were vectorized, geometrically regularized, and consolidated under a spatial-exclusivity constraint to eliminate duplicate representations. We used twenty-four predictors capturing geometric, spatial-contextual, and raster-derived spectral and texture properties. Machine Learning (ML) classifiers were trained on a development partition (Sites B-D) and rigorously validated on a spatially independent test set (Site E) excluded from hyperparameter tuning and class balancing. The experimental results demonstrate that geometric and spatial-contextual predictors using Decision Tree (DT) provide the most effective discriminatory evidence for identifying object-level boundary deformations. DT achieved an accuracy of 95.31%, an F1-score of 91.06%, and a Matthews correlation coefficient (MCC) of 0.880 on the unseen testing site. At the database level, this framework successfully identified 87.34% of erroneous footprints while maintaining 98.31% of acceptable structures, reducing the residual error proportion from 27.32% to 4.62% and improving final database purity to 95.38%. This translates into a relative error reduction of 83.09%. The findings indicate that post-segmentation object-level ML provides a highly transferable, robust mechanism for automated quality assurance in production-ready geographic information system (GIS) workflows. --- ### Diversity Matters: Distributional Feature Coverage Sample Selection for Data-Efficient Backdoor Attacks Authors: Yi Yang, Xiaoke Chen, Jinyang Huang, Feng-Qi Cui, Yu-Tong Guo, Jia-Cheng Zhao, Haiming Jin, Xiaokang Zhou, Meng Li Published: 2026-08-10 Categories: cs.CR, cs.CV arXiv: https://arxiv.org/abs/2608.09047v1 Backdoor attacks compromise training data so that a model retains clean accuracy but predicts an attacker-chosen target on triggered inputs. At very low poisoning rates, only a few samples convey the trigger--target association, making poison-sample selection critical. Existing methods typically rank candidates using per-sample scores, which can select redundant samples from similar semantic regions, and many require task-specific surrogate training. We propose Distributional Feature Coverage Sample Selection (DFCS), a training-free, trigger-agnostic method that clusters fixed pretrained features into one region per poisoning slot and selects the centroid-nearest sample from each region. A local first-order analysis relates this allocation to feature-coverage and representative-mass terms. Across BadNets and Blended attacks on CIFAR-10, Tiny-ImageNet, and Imagenette, DFCS achieves the highest mean attack success rate among seven selectors in all six dataset--attack settings, averaging $96.30\%$ and exceeding the strongest comparator in each setting by 4.60 percentage points on average while preserving clean accuracy. These results support distributional feature coverage as an effective selection principle for low-budget dirty-label backdoor attacks. --- ### Measuring the Tokenization Premium: A Cost Audit for Underserved Language Communities Authors: Avijit Roy, Proma Roy, Hrishitva Patel Published: 2026-08-10 Categories: cs.CL, cs.CY arXiv: https://arxiv.org/abs/2608.09046v1 Large language models are increasingly deployed as general-purpose educational and technical assistance systems, but their underlying infrastructure does not treat languages equally. One underexamined source of disparity is tokenization: semantically equivalent content can require substantially different token counts across languages, affecting API cost, latency, and usable context length before a model is invoked. We introduce the Tokenization Equity Audit (TEA), a reproducible benchmark for measuring tokenization premiums in technical tutoring content. TEA evaluates three widely used tokenizers, GPT-4o's o200k base, Qwen2.5-7B, and Mistral-7B, on a 120-item Python debugging corpus translated from English into Bengali, Hindi, Arabic, Tamil, and Yoruba. Bengali and Hindi serve as the primary validated cases, while the remaining languages provide exploratory cross-script and cross-family comparisons. Across this corpus, Bengali requires (1.56\times) as many GPT-4o tokens as English, reducing a nominal 128k-token context window to an effective 82k-token English-equivalent capacity for the same semantic content. With the Qwen2.5 and Mistral tokenizers, Bengali requires up to (4.5\times) the English token count. Yoruba, despite using the Latin script, exhibits the highest GPT-4o tokenization premium at (2.37\times), indicating that tokenization inequity cannot be explained by script family alone. These results demonstrate that tokenization can create measurable economic and functional barriers, highlighting the need to treat tokenization as an equity-relevant infrastructure layer for underserved language communities, particularly where educational systems depend on low-cost or offline-capable AI tools. ---