跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+816新增论文
+642其中 AI 相关
+95已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-21 12:19 · 全站最近入库:2026-11-21 12:19各信源进度 →

统计口径与数据新鲜度

统计 2026-11-20 13:03 — 2026-11-21 13:03(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-21 12:19最近发现新增版本:2026-11-21 12:19库内论文最新更新时间:2026-11-20 02:59

每分钟刷新已保存的数据。空检查不代表采集异常;新收录可能包含历史论文。AI 导读按新增版本计数。

29,158 已收录论文24,181 AI 相关8,992 有修订112 HF 社区发现

浏览论文库

时间范围跟随排序:版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常,查看状态 →

找到 24,181 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-21 12:17(北京时间)
新论文 · v1arXiv:2611.24422

Resolution as Attention for Efficient Agent Memory Perception

Siqi Xiang, Zhipeng Xu, Shengrui Liu, Zulong Chen, Chunyan Miao, Shijian Lu

原文摘要

LLM-based agents accumulate extensive interaction histories when solving complex, long-horizon tasks, making efficient memory management critical to their performance and scalability. While existing memory management methods primarily focus on what to retain or retrieve, they often overlook how much perceptual detail should be allocated to individual past interactions. We posit that effective agent memory depends not only on the information preserved, but also on the granularity at which each past interaction is perceived. Motivated by this insight, we propose Resolution as Attention (RAA), a memory perception mechanism that uses visual resolution as a means of attention allocation across interaction histories. RAA renders past interactions as images and dynamically adjusts their resolution at every step according to their relevance to the agent's upcoming decision, preserving fine-grained information for critical interactions while reducing visual-token consumption for less relevant ones. To adaptively determine the perceptual fidelity of each interaction, we train a lightweight relevance estimator with teacher-model guidance and further optimize it to preserve the agent's decision consistency at a reduced visual-token cost. Experiments across multiple tasks and backbones show that RAA reduces token consumption by over 50% without compromising task performance. Moreover, RAA can be integrated as a plug-in perception layer with existing memory management methods, yielding further token savings and revealing adaptive perception as a complementary dimension of agent memory.

cs.AI · cs.CV

新论文 · v1arXiv:2611.24419

PhysResponse: Probing the Physics of Video Generation with Controlled Interventions

Yuhan Li, Shuran Zhang, Haixin Tian, Baonuo Rao, Weipeng Li, Yujia Hou, Zhenyu Wang, Zijun Wang 等 15 位作者

原文摘要

Video generation models are increasingly viewed as world simulators, which requires not only plausible motion but dynamics that respond correctly to specified physical parameters. Existing models, datasets and benchmarks, however, largely target physical plausibility, leaving this physical parameter responsiveness insufficiently explored. We introduce PhysResponse, a benchmark that measures it at two levels: scene-conditioned simulator inversion recovers the effective gravity, friction and initial speed of generated motion, and event evaluation tests outcomes near and across critical transitions. Evaluating ten image-to-video models, we find that even the lowest parameter errors exceed those of physically correct videos by over an order of magnitude, that models rarely reproduce both outcomes of a transition pair, and that the two levels expose complementary failures that visual metrics such as PSNR miss. To supply the missing training signal, we construct 72,000 simulated videos of 20 multi-object interaction types with controlled parameters; fine-tuning on them lowers parameter errors, even on held-out interaction types, and transfers to real videos and an external physics benchmark. Continued training with a numerical encoder further lowers all three errors on held-out types. PhysResponse provides a foundation for world simulators that follow physical conditions, not just appearances.

cs.CV

版本更新 · v2arXiv:2606.00740

Learning Binary Codes with Spiking Neural Networks for Cross-Modal Hashing Retrieval

Yukuan Zhang, Jiarui Zhao, Shangqing Nie, Jin Kuang, Dong Liu, Shengsheng Wang

原文摘要

Cross-modal hashing (CMH) uses compact binary codes for efficient image-text retrieval. Most methods learn image-text relationships in a continuous space and binarize the representation only at the end. Semantic learning and bit decisions thus remain largely separate. This work addresses two key questions concerning hash-bit generation: \textbf{(1) How to determine hash bits?} We leverage the discrete nature of spiking activity and determine each bit through competition between positive and negative spikes accumulated over time, rather than through a single firing event or subsequent thresholding of continuous representations. \textbf{(2) How to generate such hash bits?} We propose an unsupervised spiking framework for cross-modal hashing, in which image and text jointly contribute to hash-bit generation through temporal interaction. An activation constraint further prevents unsupported decisions caused by silent channel pairs. Extensive experiments on three benchmarks demonstrate the effectiveness of our approach and support temporal spike competition as a design principle for hash-code generation.

cs.IR · cs.MM

新论文 · v1arXiv:2611.24413

MINT: Meeting-time Indicators for Truncation in Multi-Step Off-Policy RL

Seungyub Han, Taehyun Cho, Dohyeong Kim, Kyungjae Lee, Jungwoo Lee

原文摘要

Many multi-step off-policy temporal-difference methods handle policy mismatch through products of per-step trace coefficients, leading to a trade-off between variance control and long-horizon signal. The maximum trace length consequently becomes a sensitive hyperparameter. We propose MINT (Meeting-time INdicators for Truncation), which replaces these products with a binary stopping indicator: one until coupled behavior and target trajectories first meet, zero thereafter. At the true value function, the indicator's idempotence yields a variance bound governed by meeting-time survival. For practical implementation, a single sampled action provides a model-free surrogate for the stopping rule, requiring no transition model or mixing-time estimate. We derive separate sufficient conditions for contraction of the state-coupled operator and its practical surrogate under fixed-policy evaluation. On MuJoCo and DeepMind Control tasks, MINT outperforms the reported multi-step baselines; the main MuJoCo comparison uses a common cap without per-task cap-length tuning. We also demonstrate the effectiveness of MINT on Atari 100k, RWRL, and OGBench experiments; MINT adapts the backup horizon to reduce target variability relative to long fixed-horizon backups across tasks and dynamics.

cs.LG · cs.AI

新论文 · v1arXiv:2611.24410

Why Multi-Source Audio Fails to Compose: A Geometric Diagnosis and Its Remedy

Anurag Singh, Prajwal Singh, Shanmuganathan Raman

原文摘要

Multi-source audio-to-image generation exhibits consistent compositional failures when audio tokens occupy geometrically displaced, semantically unanchored positions in the CLIP text-token space, where the diffusion U-Net lacks the prior to reliably attend to them or ground them in distinct spatial regions. We formalize this as the geometric displacement phenomenon and expose it through a two-level clustering analysis comparing BEATs and CLAP embeddings at the raw encoder and projected CLIP token levels. Three geometric properties emerge as important: intra-class compactness, angular alignment with the CLIP text manifold, and inter-token spatial separation. We show that CLAP, whose contrastive language-audio pretraining aligns embeddings with natural language semantics, substantially outperforms BEATs on all three properties, improving the silhouette score by more than 3x and the Audio-Image Content score as well. Building on this analysis, we introduce a multi-token extension that requires no diffusion-backbone retraining and assigns each audio source to a distinct placeholder token in the CLIP tokenizer, complemented by a mixed inference-time objective that combines Attend-and-Excite with an A* spatial separation loss to resolve residual cross-token interference. For evaluation, we construct multi-audio compositional benchmarks from VGGSound (VGG2, VGG3, and VGGN) and extend six standard audio-image metrics to the multi-source setting. Across five benchmarks and four strong baselines, our method achieves strong results on the metrics most directly tied to compositional faithfulness, with AIC improvements of up to 2.25x over the strongest baseline and FID reduced by nearly 2x on two-source generation. We propose geometric displacement as a diagnostic framework for cross-modal generation systems that project learned embeddings into a pretrained contrastive space.

cs.CV · cs.SD

新论文 · v1arXiv:2611.24409

Risk-Triggered Search: Behavior-Preserving Safe Control

Xinyu Deng, Bohao Lan, Kunpeng Li, Mingkai Yuan, Weihang Pan, Binbin Lin

原文摘要

Pre-trained visual world models generate multi-step action plans from current and goal observations, yet terminal-proximity objectives fail to prevent forbidden-zone traversal or rigid-body collision sliding. Purely visual interfaces may lack analytical dynamics, structured states, or explicit collision geometry, requiring internal prediction of action-conditioned process risk. We propose Risk-Triggered Search (RTS) for frozen Joint-Embedding Predictive Architecture (JEPA) world models. Rather than relying on static state features that cannot distinguish candidate plans sharing an observation, RTS unrolls action-conditioned future latent prefixes in the frozen world model and trains a lightweight risk head with process-violation supervision. To intervene selectively while preserving safe nominal behavior, RTS uses the same predicted prefix-risk criterion as both the search trigger gate and hard feasibility gate: safe nominal plans execute directly, whereas alarms trigger local Cross-Entropy Method (CEM) search to minimize nominal deviation within the feasible set. To prevent isotropic Euclidean minimal modification from shrinking obstacle-directed actions into zero-amplitude stalling, we introduce an amplitude-preserving metric $d_{\mathrm{AP}}$ that retains tangential progression while maintaining a unique zero at the nominal plan. Under unified gating and this unique-zero condition, bypassing safe-state search yields closed-loop trajectories strictly identical to every-step same-kernel search. Evaluated under a process-safety benchmark and Safe Arrival Rate (SAR) metric across five tasks, RTS improves SAR over Direct by 12.67--41.67 percentage points on navigation and articulated control and +2.08/+4.17 pp on Cube/PushT boundary cohorts, maintaining 100% action identity with every-step search while cutting controller time by 17.71%--70.41% across 1,968 paired trajectories.

cs.AI

新论文 · v1arXiv:2611.24408

StrideWorld: Benchmarking Scene Rigidity and Action Consistency in Interactive Video World Model

Chuang Wang, Rui Nie, Jie Song, Junyuan Gao, Jiahe Song, Haitao Zhou, Boyu Zhu, Xing Li 等 12 位作者

原文摘要

Interactive video world models turn user commands into the exploration of a generated scene, which requires the scene to keep its 3D structure as the camera moves and each action to keep a stable effect. Existing benchmarks mostly evaluate these properties against references fitted to the video under test, either a scene reconstructed from its own frames or a camera trajectory aligned to the command path with a free scale. Such references can absorb gradual scene deformation and changes in pace and route length, so these errors can pass undetected. We therefore introduce StrideWorld, a benchmark whose references are fixed before the motion they judge. It evaluates scene rigidity against the ground-truth geometry of the start view, and action consistency against the stride of each action, the camera motion of one command. The stride is measured on the model's first steps and then used to test the steps that follow, so each model is judged only against its own steps. Its measurements are validated on rendered videos with known camera motion, and its scores agree with human judgments. Across \nvideos{} videos generated by \nmodels{} open models from $140$ start views, all models but one deform their scenes as the camera moves, with offsets up to $7.3$ times those of a rigid rendering along the same camera path. Most models keep a steady pace when a single action is held. Along composed routes, several models often stall, and most completed routes end away from the positions predicted by each model's own strides. Only the two models that scale commands by the estimated scene depth bring most routes close to these positions. Fine-tuning two models on a few rendered clips removes most of their mid-route stalls. Code, data, and a leaderboard are publicly available at https://github.com/chuangwang123/strideworld.

cs.CV

版本更新 · v2arXiv:2609.28559

Who Is Behind the Harness? Verifying Model Identity in Coding Agents through Behavioral Fingerprinting

Chuyi Wang, Xiaohui Xie, Tongze Wang, Fangchen Luo, Yong Cui

原文摘要

LLMs increasingly operate through coding-agent harnesses that inspect repositories, invoke tools, and modify files. Substituting the model behind such an agent can therefore change security-relevant decisions, including whether it verifies changes or recovers safely from failures. Existing LLM fingerprints largely infer identity from direct text or token distributions. In coding agents, these signals are mediated by system instructions, controller logic, tools, and execution feedback, limiting their transfer. We present LIDAR (LLM Identification from Decisions and Actions at Runtime), an active black-box fingerprinting method for coding-agent execution. Three coding probe pairs expose post-edit verification, transient-failure recovery, and specification--test conflict resolution under controlled changes. LIDAR represents the resulting trajectories with complementary instance-level and distribution-level features and compares them with clean references using a lightweight probabilistic identifier. It requires no access to model weights, logits, or provider internals. We evaluate 24 models from eight families on OpenCode, OpenHands, and mini-swe-agent, fitting separate references for each harness. Closed-set Top-1 accuracy ranges from 91.39\% to 98.89\% across the three harnesses, exceeding all three baselines in each setting. When the audit also includes unregistered models from the same families, balanced accuracy ranges from 79.58\% to 80.42\%. These results show that agent execution behavior provides model-identity evidence beyond final outputs.

cs.CR · cs.AI · cs.SE

新论文 · v1arXiv:2611.24406

CaDO: Formal Verification-Guided Multi-Agent Framework for Causal Dataset Repair and Generation

Songli Yu, Qi Luo

原文摘要

High-quality causal reasoning datasets are essential for developing models with reliable causal understanding, particularly in specialized domains. However, existing datasets frequently suffer from incorrect annotations and structural imbalances, which can lead to biased predictions and limited generalization. To address this issue, we propose CaDO (Causal-Engine-Guided Multi-Agent Data Optimization), a multi-agent framework for automated diagnosis and repair of causal reasoning datasets. Our framework consists of three key stages: formal causal reasoning, multi-level diagnosis, and automated data repair and generation. First, a Formalizer Agent translates natural language questions into formal causal queries to minimize semantic deviations and make retrieval and verification more explicit and tractable. Second, a dual-level diagnosis mechanism employs independent Critic Agents to identify sample-level errors while analyzing query-type coverage and label distributions to detect dataset-level structural imbalances. Finally, the framework employs reusable repair operators grounded in formal causal semantics to correct erroneous samples, while Explorer and Expander Agent collaboratively produce diverse and semantically consistent new samples. Experimental results show that CaDO significantly outperforms the strongest of five data cleaning baselines in 16 out of 18 experimental settings, improving downstream model accuracy by up to 9.1%. Moreover, at least 95% of the samples flagged by CaDO are defective, and on CLadder, CaDO incorrectly modifies 92.2% fewer correct labels than an LLM-based judge, demonstrating its effectiveness and reliability. Our code and data are publicly available at https://github.com/thisissyu/cado.

cs.AI · cs.MA

版本更新 · v2arXiv:2608.21274

Recommendation Quality and the Concentration of Consumption: Experimental Evidence from Netflix

Guy Aridor, Winston Chou, Nathan Kallus, Antoine Scheid, Allen Tren, Kevin Zielincki

原文摘要

We study an experiment with 8.5 million users on Netflix's recommender system to measure how improvements in recommendation technology affect the set of products that get consumed. Improvements increase total consumption and users' reliance on recommendations while diffusing recommendations and consumption away from the most popular titles ("superstars") toward a larger number of moderately popular titles ("middle-tail"), with minimal effects on the most niche titles ("long-tail"). Our results challenge the notion that recommender systems polarize consumption -- raising the consumption shares of the head and tail at the expense of the middle -- and suggest that the returns to investing in middle-tail products grow as algorithms improve and platforms scale.

econ.GN · cs.IR

版本更新 · v3arXiv:2609.20151

Ischemic Stroke Segmentation and Net Water Uptake Quantification on Multicenter Non-Contrast CT Using Supervised Target-Domain Adaptation

Linus Britt, Maximilian Nielsen, Susan Klapproth, Andre Kemmling, Michael H. Lev, Gabriel Broocks, Rene Werner, Thilo Sentker

原文摘要

Quantitative assessment of infarct hypodensity on non-contrast computed tomography (NCCT), including net water uptake (NWU), requires manual or semi-manual lesion delineation, often guided by CT perfusion or diffusion-weighted MRI, limiting clinical applicability. Automated NCCT segmentation could enable such biomarker extraction but remains challenging across heterogeneous multicenter data. This study developed and externally tested a deep learning framework for ischemic stroke segmentation and NWU quantification on NCCT. In this retrospective multicenter study of 801 patients from four datasets, an nnU-Net-based model with foundation-model initialization, extended by modules comparing local features with the contralateral hemisphere, was trained on NCCT scans from the University Medical Center Hamburg-Eppendorf and the Acute Ischemic Stroke Dataset. The model was fine-tuned on target-domain subsets of the Boston (n=11) and ISLES (n=75) datasets, with evaluation on held-out cases. Segmentations and NWU were compared with expert references. Internally, mean Dice increased from 0.53 (conventional nnU-Net) to 0.57 (proposed approach). After fine-tuning, the proposed approach achieved median Dice of 0.56 for acute lesion segmentation (Boston) and 0.26 for post-treatment final-infarct prediction (ISLES), where small lesions predominated. For lesions >=30 mL, median Dice increased to 0.67 and 0.57, respectively. On the Boston dataset, automated and reference NWU agreed (concordance correlation coefficient 0.93, mean absolute error 1.24 percentage points). Contralateral feature comparison and target-domain adaptation supported NCCT-only infarct segmentation across heterogeneous external cohorts, although performance was lower for post-treatment final-infarct prediction. The approach enabled low-error NWU quantification from baseline NCCT without advanced imaging or manual annotation.

cs.CV

新论文 · v1arXiv:2611.24404

RAPID: Reconstruction Accelerated by Polarization-Informed Depth for 2D Gaussian Splatting

Daisong Ouyang, Linlin Xia, Liangkun Yan, Song Gu, Shedong Zhao, Xin Dai, Yi Zheng

原文摘要

Efficient and accurate 3D surface reconstruction from multi-view images remains a fundamental challenge in Gaussian splatting. Many existing methods rely on sparse geometric initialization and iterative densification, resulting in substantial optimization overhead and difficulties in balancing rendering quality and geometric accuracy, particularly in reflective and weakly textured regions. To address these limitations, we propose RAPID, a geometry-first framework that uses multiview polarization cues to recover dense surface priors. We integrate surface normals into depth maps using sparse geometric anchors. These depth maps are then fused across views. Curvature-adaptive sampling and oriented Gaussian initialization then support efficient optimization. Experiments on the PISR dataset demonstrate an approximately \(5\times\) speedup over PhyGaP, which is a state-of-the-art Gaussian-based reconstruction method. RAPID reduces the average optimization time from approximately 21 minutes (PhyGaP) to 4 minutes while maintaining competitive novel-view synthesis quality and mesh reconstruction accuracy. These results demonstrate the effectiveness of RAPID in accelerating Gaussian optimization from polarimetric multi-view inputs.

cs.CV

版本更新 · v2arXiv:2608.04804

Scrouting: Cost-Aware Routing of Coding Agents by Scouting the Repository First

Ishaan Bhola, Adithyan Krishnan, Mukunda NS, Dhruv Jaglan

原文摘要

Frontier language models can resolve repository-level software issues, but each attempt is expensive, and existing routers select a model from the issue text alone. We present SuperScout, which routes after scouting the repository: a 7B searcher, SuperScout-7B, first explores the repository and produces a structured handoff whose reproduction claims are sandbox-verified, with false claims stripped before delivery. The searcher's hidden states, together with the task text, then feed a resume-based router that dispatches the task to one of four frontier fixers. Adding a new fixer requires no retraining. On the full Python slice of SWE-bench Pro (266 tasks) under the benchmark's official capped budget tier, SuperScout matches the best single model's solve rate (159 of 266 for SuperScout, 158 for the best model) at about a fifth of the total cost per solve, and the reported configuration sits above the random traffic-splitting baseline. A no-router ablation, always the cheapest fixer with the handoff, ties the routed system on this benchmark, so the handoff rather than the routing decision carries the result. A paired calibration study points to the mechanism: the handoff appears to redistribute rather than add solving ability, lifting the three cheaper fixers while slightly hurting the strongest, though at $N=99$ the per-fixer effects are directional only; the searcher's hidden states improve cost routing on the calibration labels while the handoff's own text does not. The searcher's compute adds less than half a cent of GPU time per task.

cs.SE · cs.AI

版本更新 · v3arXiv:2604.09870

Reading Human Preference from Looped Transformer Internal States

Jan Kirin

原文摘要

Show a frozen language model two replies to the same request, one of which a person preferred. Can a small probe on the model's hidden states tell which? On Ouro-2.6B-Thinking, a model that runs its layers four times per token, and Anthropic's HH-RLHF data, the answer is yes, but only faintly. A linear probe picks the preferred reply 61.5% of the time, a small nonlinear network that scores each reply separately 64.9%, and the evaluator trained for version 1 reaches 64.7% on the test split: clearly better than simple rules such as picking the longer reply, but weak. Whether preference is stored as a comparison between the two replies or as a property of each is still open. A linear probe cannot tell, and a network that reads both replies together has an estimated edge of only 0.7 points (95% CI 0.0 to 1.4) over one that scores each reply separately. These numbers hold under three controls: scoring every pair in both orders, splitting the data by pair, and, for the probes, keeping the end of each conversation, where the two replies differ. Version 1's 95.2% and 84.5% were obtained without the first two and do not hold; version 2's corrected relational-probe accuracy, obtained without the third, was four to five points too low. This version replaces both.

cs.LG · cs.AI

新论文 · v1arXiv:2611.24401

Reliable Tree Recruitment from Scarce Annotations for Forest Inventory Expansion

Luiz Fernando Souza Luz, Pedro J. Soto Vega, Matheus Pinheiro Ferreira, Laura Elena Cué la Rosa, Hudson Franklin Pessoa Veras, Ernandes Macedo da Cunha Neto, Evandro Orfanó Figueiredo, Ana Paula Dalla Corte 等 18 位作者

原文摘要

Species-level inventories of individual tree crowns (ITCs) from UAV imagery are limited by annotation cost: in tropical forests, many species are represented by only a handful of field-validated crowns. We propose a reliable tree recruitment framework that expands a small annotated inventory by iteratively self-training a multi-task DeepLabV3 network, combining semantic segmentation with a crown-distance regression head, on RGB UAV orthomosaics. At each iteration, candidate crowns enter the training inventory only if they pass a confidence filter on class probability and distance map and a morphology filter on crown area and solidity, with at most five new crowns per species, keeping recruited samples consistent with the reference inventory. We evaluate the framework at two Brazilian sites: an Amazonian rainforest in Acre (17 species, 355 annotated crowns, 4 cm GSD) and a forest site in the Cerrado of Maranhão (4 species, 51 annotated crowns, 1.8 cm GSD). Over 20 iterations, the number of retained crowns grew from about 500 to 960 in Acre and from about 700 to over 3100 in Maranhão, while crown-level F1 exceeded single- and multi-task DeepLabV3 baselines (53.5% vs. 50.0% and 45.2%; 92.2% vs. 79.7% and 83.6%). The framework offers a low-cost route to expanding species inventories with consumer-grade UAVs and limited fieldwork.

cs.CV

新论文 · v1arXiv:2611.24395

RSIBench-Infra: Can Agents Optimize Their Own Training and Inference Infrastructure at Production Scale?

Hanlin Wu, Yifei Wang, Zhicheng Cai, Huan-ang Gao, Wei-Ying Ma, Ya-Qin Zhang, Jingjing Liu, Hao Zhou

原文摘要

Improving the efficiency of training and inference infrastructure is a practical component of recursive self-improvement. Existing benchmarks for agent-driven optimization largely focus on individual software layers such as kernel, leaving unclear whether agents can deliver end-to-end performance gains under production workloads. We introduce RSIBench-Infra, a benchmark of 20 training and inference tasks spanning diverse model families and modalities, with deployments of up to 32 accelerators on NVIDIA GPUs and Ascend NPUs. Agents start from runnable baselines derived from official recipes and can modify the software stack across engine, kernel, and instruction levels. An OCI image-layer submission protocol captures these changes for reproducible deployment, while independent evaluation measures end-to-end performance under task-specific quality constraints. Across all task results, 66.2% exceed $1.10\times$ validated speedup. These results demonstrate that current agents can improve production-scale training and inference performance, but identifying effective optimization strategies and consistently translating local improvements into end-to-end gains remain challenging.

cs.LG

版本更新 · v4arXiv:2603.22665

Improving LLM Final Representations with Inter-Layer Geometry

Tom Ulanovski, Eyal Blyachman, Maya Bechler-Speicher

原文摘要

The standard in LLM-based prediction is to use the final-layer representation as the input to a downstream predictor. However, intermediate layers may encode complementary task-relevant signals. Existing approaches therefore either search for the best layer for each task or apply expensive attention-based mechanisms to learn inter-layer aggregation. In this work, we first show that such complexity is unnecessary: a lightweight Graph Neural Network over a fully connected graph of LLM layers is more efficient and achieves significantly stronger predictive performance than existing approaches. We then introduce the Cayley-Encoder, which further improves both efficiency and predictive performance by replacing the fully connected graph with a Cayley graph over SL(2, Zn). These Cayley graphs provide a mathematically grounded topology that is sparse, regular by construction, and has low diameter. This enables effective communication across layers while constraining the aggregation structure and reducing the risk of GNN overfitting. In an evaluation of Cayley-Encoder across 13 tasks and 13 LLMs, Cayley-Encoder consistently outperforms baselines, achieving improvements of up to 48 percentage points in accuracy, while introducing at most 0.1% additional parameters relative to the LLM size. We further show that Cayley-Encoder is effective in few-shot regimes. Finally, we show that Cayley-Encoder outperforms LoRA fine-tuning while operating on the frozen LLM. We conclude with an explainability analysis showing that multiple layers contribute meaningfully to the final prediction, supporting our hypothesis.

cs.CL · cs.LG

新论文 · v1arXiv:2611.24394

Temporal Rank in Recurrent Neural Networks

Leo Dobson, Anton Ragni

原文摘要

Recurrent Neural Networks (RNNs) are a natural framework for temporal modelling, yet are often dismissed due to their sequential, non-parallelisable computation and difficulty in capturing long-range dependencies, typically attributed to vanishing and exploding gradients. We identify a distinct and under-explored limitation: an information bottleneck induced by rank-constrained hidden state dynamics, which limits the number of independent input directions that can be preserved over time, as reflected in the rank of the input-to-state mapping. We show that this constraint restricts temporal capacity independently of gradient stability. To address this, we introduce structured high-dimensional state dynamics that preserve temporal separability and prevent saturation of the input-to-state Jacobian over time. On controlled sequence tasks, the resulting model exhibits rapid learning despite remaining strictly within a standard RNN formulation. Our results demonstrate that RNN performance is governed by the evolution and preservation of temporal rank, and suggest that mitigating gradient pathologies alone is insufficient without explicitly maintaining temporal capacity.

cs.LG · cs.AI

版本更新 · v2arXiv:2610.04433

What Does a Harness Buy? Tokens, Mostly

Yangze Liu, Zhongyi Han

原文摘要

A coding agent is a language model inside a harness: the system prompt, the tools, and the context management that let a chat model work in a repository. Most of the engineering on these agents goes into the harness, and vendors and leaderboards credit it with score gains. Earlier studies have fixed the model and swapped the harness, but have not measured whether the swap changes the pass rate more than running the same harness again. We run seven models through three production harnesses, Claude Code, mini-SWE-agent, and OpenCode, on SWE-bench Verified and two benchmarks of other kinds, and rerun them unchanged. No harness leads consistently across models. Where the data can bound the difference between two harnesses, the bound is 2 to 5 points, and a harness swap changes about as many task outcomes as a rerun. The exceptions are harness effects in particular pairings of model and harness, whose direction changes with the model: OpenCode trails by 8 to 9 points on one model and leads by 14 on another. Cost differences between harnesses also depend on the model, from under 10% to 3 to 4 times. Which harness is better therefore has no answer apart from the model, and a harness difference should be reported next to a rerun.

cs.AI · cs.LG · cs.SE

版本更新 · v8arXiv:2507.06174

Design and Experimental Validation of Sensorless 4-Channel Bilateral Teleoperation for Low-Cost Manipulators

Koki Yamane, Yunhan Li, Masashi Konosu, Koki Inami, Junji Oaki, Toshiaki Tsuji, Sho Sakaino

原文摘要

Teleoperation of low-cost manipulators is attracting increasing attention as a practical means of collecting demonstration data for imitation learning. However, most existing low-cost systems rely on unilateral position control without force feedback, while implementing force-feedback bilateral teleoperation is difficult because low-cost manipulators typically have low-resolution encoders and no joint torque sensors. This paper presents a sensorless 4-channel bilateral teleoperation framework that integrates identified nonlinear dynamics compensation with a disturbance-observer-based velocity and external-force estimation scheme. By interpreting the observer structure in the frequency domain, we clarify the coupling between the velocity- and external-force-estimation bandwidths and derive practical tuning guidelines based on the damping ratio and a single cutoff frequency. Real-robot experiments, including force-sensor comparison and teleoperation tasks, demonstrate that the proposed framework provides practically useful force estimates and enables stable teleoperation in high-speed and contact-rich scenarios under low-cost hardware constraints. As an application, imitation-learning experiments demonstrate that incorporating estimated force information into demonstrations improves task success rates in the tested contact-rich manipulation tasks.

cs.RO · cs.AI · eess.SY

版本更新 · v2arXiv:2611.09373

Calibrated Uncertainty Quantification on TDC ADMET: What Can Be Resolved Across 22 Datasets?

Laurits Fredsgaard, Michael Riis Andersen, Mikkel N. Schmidt

原文摘要

Absorption, distribution, metabolism, excretion, and toxicity (ADMET) models often rely on small, heterogeneous datasets, yet their predictions may determine which compounds proceed to experimental testing. Useful evaluations must therefore assess both predictive performance and the quality of reported probabilities or intervals. We ask which common uncertainty-quantification methods perform best on the 22 Therapeutics Data Commons (TDC) ADMET datasets and what this benchmark can resolve. We compare eleven calibrated pipelines using nested five-fold cross-validation across ten chemical partitions and five model seeds. Ensemble Monte Carlo (MC) dropout has the best average rank for TDC performance, Brier score, and 90% interval score. For TDC performance, however, it is not separated from the other ensemble pipelines; for Brier and interval score, LightGBM is also not ruled out as best. LightGBM wins more individual datasets than any other method for TDC performance and Brier score, but performs poorly on several endpoints. Smaller evaluations can identify a different leader and often fail to recover the broader comparison. These results make LightGBM and an ensemble neural model complementary baselines. Comparative claims should be checked across multiple chemical partitions using statistical analyses matched to the question being asked.

cs.LG

版本更新 · v3arXiv:2609.07441

TabBench-Bio: A Living Benchmark for Machine Learning on High-Dimensional Biomedical Tables

Jules Kreuer, Sofiane Ouaari, Julia Hellmig, Julius Braitinger, Nico Pfeifer

原文摘要

Biomedical tables often combine thousands of measured variables with only tens or hundreds of labelled samples, a regime that is poorly represented in general-purpose tabular benchmarks. We introduce TabBench-Bio, a living and interactive benchmark of 43 biomedical datasets spanning multiple domains. Under a shared cross-validation protocol, we compare classical estimators, neural networks, and tabular foundation models across 28 feature-by-sample operating points. At the reference cell of 10,000 features and 100 training samples, RealTabPFN 2.5 has the highest point estimate, closely followed by TabPFN 3 and Logistic Regression. A paired bootstrap over the target pool separates RealTabPFN 2.5 from TabPFN 3 by 87 Elo (95\% interval [43, 129]). Tabular foundation models generally occupy the leading ranks, while the strongest configuration depends on the operating point and biomedical modality. The AutoML framework AutoGluon, using its one-hour "extreme" preset, is configured as a separate resource-intensive reference and reported here at the reference and full cell. Fold-level predictions, run status, and deterministic aggregations make every reported result reproducible and reusable. The benchmark is open to contributions of new biomedical datasets. The interactive leaderboard is available at https://tabbench-bio.eu.

cs.LG · cs.AI

版本更新 · v2arXiv:2608.28008

Visual Token Coding for Video Multimodal Large Language Models

Chenxin Fang, Tao Chen, JunChao You, Jun Peng, Yiyi Zhou, Rongrong Ji

原文摘要

In this paper, we propose a new token compression paradigm for video Multimodal Large Language Models (MLLMs), termed Visual Token Coding ($VTC$). Inspired by classical video coding principles, e.g., HEVC, VTC performs structured compression by predicting the I/P frames of a video and measuring their frame-wise residuals to estimate token redundancy. Based on this base paradigm, we further extend VTC with a set of dynamic budgeting designs towards more cost-effective token coding, termed $VTC_{dy}$. To validate VTC and $VTC_{dy}$, we apply them to three MLLMs and conduct experiments on a set of video understanding benchmarks. The experimental results show that as a base paradigm, VTC can achieve an average performance retention of 98.4% on Qwen3-VL under a 50% token compression ratio, and the enhanced $VTC_{dy}$ can further improve the retention to 100.3%, outperforming a set of compared methods. Moreover, as a plug-and-play design, VTC requires no additional tuning of MLLMs for token coding. Our code is given in the supplementary materials.

cs.CV

版本更新 · v3arXiv:2510.10605

Hyperparameter Optimization via Budget Allocation in Lipschitz Spaces

MohammadHossein Bateni, Hossein Esfandiari, Samira HosseinGhorban, Alireza Mirrokni, Radin Shahdaei

原文摘要

Hyperparameter optimization and model selection often require allocating a limited evaluation budget across many candidate configurations, where performance improves with additional training or resources. We study this setting when each configuration has an unknown performance curve over budget. Similarity between configurations induces similarity between their curves, capturing geometric structure in the search space. We formalize the resulting decision problem as Unknown Value Probing and propose budget allocation strategies that combine principled exploration of the configuration space with value-informed concentration on regions supported by observed evidence. We prove worst-case approximation guarantees for our approach and show these guarantees are near-optimal by matching corresponding hardness barriers. Extensive experiments on large benchmark suites spanning more than 250 experimental settings demonstrate consistent gains over strong baselines in both anytime behavior and final incumbent performance.

cs.LG

新论文 · v1arXiv:2611.24384

Enhancing Active Learning Performance with Initial Data Selection Based on Foundation Models

Mohammad Mirzaee, Mehrshad Fallah Estalkhzir, Mohammad Reza Mohammadi

原文摘要

In active object detection, selecting the initial labeled pool is essential, especially in complex datasets with expensive labeling and low resource budget. This study presents an active learning framework designed to optimize the initial training cycle of active learning for the object detection task. Our approach diverges from traditional methods by utilizing feature-based clustering from the YOLO-World model and refining images based on objectness scores rather than relying on uncertainty sampling in the initial cycle. This approach focuses on finding samples with various features and emphasizes training in areas where the foundation model is most confident. Doing so helps the model better understand different types of objects. For subsequent cycles, we employ a consistency-based sampling technique to maintain the robustness of the training process. To evaluate our method, we use two complex datasets called Aquarium and NWPU VHR-10, which are well known for their diverse object sizes and challenging conditions. In the NWPU dataset, our method increases mAP50 by an average of 6.6% in image-based evaluation. Moreover, the image refinement method shows an improvement of approximately 11% in object-based evaluation. The results show improvements in model performance and labeling time, demonstrating the advantages of using foundation models in early cycles of active learning. An implementation of the algorithm is available at https://github.com/Mhrshd-fllh/ALBoost-FM

cs.CV · cs.AI

版本更新 · v2arXiv:2606.21453

Robustness of Text-Only Adapted Speech LLMs for Spoken Semantic Parsing

Youngwon Choi, Donghyuk Jung, Hyeonyu Kim, Taeyoun Kwon, Myeongkyun Cho, Moon Hwan Kim

原文摘要

Large language models (LLMs) are increasingly used to generate structured outputs from natural language instructions, including semantic frames and function calls. When the input is speech, this task becomes spoken semantic parsing, which can be performed using either Speech LLMs or ASR-LLM cascades. Although both approaches can be adapted with text-only task supervision, their relative robustness to acoustic degradation remains unclear. We compare these approaches under matched text-form supervision using 3B and 7B models across three datasets, evaluated on clean speech and babble noise at five SNR levels. Results show that Speech LLMs generally achieve better performance in command understanding, intent prediction, and function calling, particularly under severe noise, while improvements in fine-grained entity prediction are less consistent. An additional evaluation shows that Speech LLMs achieve lower average inference time per sample than ASR-LLM cascades at both model scales.

cs.HC · cs.AI · cs.SD · eess.AS

新论文 · v1arXiv:2611.24383

What Majority Labels Hide: Annotator Disagreement and Model Errors in Multi-Dialect Arabic Sexism Detection

Alaa Bazaid, Arkaitz Zubiaga

原文摘要

Judgements of sexism are subjective and may vary across communities, yet sexism detection datasets usually reduce annotator judgements to a single majority label, obscuring disagreement and how models behave on contested cases. We introduce MuDAS, a dataset of YouTube comments in Saudi, Egyptian and Moroccan Arabic, each labelled by three native speakers of its dialect, with the individual annotations preserved. We use it to evaluate Arabic encoders fine-tuned with hard-label, soft-label and perspectivist objectives, GPT-5.6 Terra, and four Arabic-centric large language models. Model errors are more frequent on comments with annotator disagreement. Despite relatively close Macro-F1 scores, GPT-5.6 Terra and the fine-tuned encoders exhibit contrasting error patterns: GPT-5.6 Terra, even with the annotators' definition of sexism, over-classifies gender-related content such as religious advice on modesty, whereas the encoders miss implicit and prescriptive sexism. The Arabic-centric models also over-predict sexism and remain below the encoders. Disagreement-aware learning preserves Macro-F1, but soft labels improve alignment with annotators on contested comments at the expense of unanimous ones, and neither paradigm reliably identifies contested comments. Lower performance on Egyptian Arabic, clearest for the large language models, is not fully explained by the examined measures of test-set composition and dialectness. Majority-label scores can therefore hide contrasting error patterns that matter for content moderation.

cs.CL · cs.CY · cs.SI

新论文 · v1arXiv:2611.24381

How to Turn an Attention Neural Network into an Interpretable Hard-Logic Predicate Circuit: Truth-Table Predicate Extraction and State-Machine Merging under Ship-of-Theseus Replacement

Yuchen Li, Zongqi Fan, Xintong Li, Nguyen H. Tran, Ken-Tye Yong

原文摘要

We convert a trained attention network as a whole, attention included, into a hard-logic predicate circuit that runs in its place. Every unit of the circuit is a 0/1 predicate on tokens, and every map between predicates is a minimised truth table. Short tables, such as the product of a query code and a key code, are minimised exactly, and wide feature-to-feature transitions become implication rules by noise-tolerant minimisation. Blocks are replaced one at a time from the output down, in the manner of the Ship of Theseus, and after each replacement the remaining soft part re-adapts to the frozen circuit. Attention selection takes two forms, a row-wise comparison of integer scores (order selection) and a learned rule (rule selection). Both are fitted on the ordinary training set alone, and only on the examples that the teacher classifies correctly, so that the circuit imitates the teacher where the teacher is right. On TweetEval-emotion, the complete circuit of DistilGPT2 reaches a test gold accuracy of 66.5 $\pm$ 1.3% with order selection and 57.4 $\pm$ 2.5% with rule selection over three seeds, against 74.7% for the teacher and 64.0% for a bag of word pieces fitted on the same tweets, and that of BERT-base reaches 70.0 $\pm$ 0.5% with order selection and 63.7 $\pm$ 1.2% with rule selection against 77.8%. The circuit needs no multiplier and no weight traffic, and its exported hardware description reproduces it bit for bit. By a synthesis-based estimate it classifies one tweet in about 1-2 $μ$s (DistilGPT2) and 3-5 $μ$s (BERT) on an FPGA card and in 5-9 $μ$s on a palm-sized board whose module draws 7.5-15 W, two to three orders of magnitude faster than the network on a GPU at batch 1 (measured, 0.9-2.6 ms for DistilGPT2 including an optimised CUDA-graph baseline), at a cost of 8.5 (DistilGPT2) and 8.6 (BERT) points of gold accuracy for the synthesised seed-0 circuits.

cs.LG · cs.AR

新论文 · v1arXiv:2611.24380

Where Should a Diffusion Transformer Meet Its Teacher?

Zhuoqi Hu, Shiwei Li, Haozhao Wang, Xiandi Luo, Yihao Ouyang, Wenting Shen, Ang Wang, Jiamang Wang 等 10 位作者

原文摘要

Representation Alignment (REPA) significantly accelerates Diffusion Transformer training by guiding intermediate student features with a pretrained visual teacher, yet existing methods typically attach supervision to a single fixed layer. As student representations evolve, however, we find that the most effective layer for alignment varies across training stages. To track this evolving layer suitability online without costly search, we measure gradient compatibility between the flow-matching and alignment objectives, and propose AdaREPA, an online routing framework that adaptively selects the supervision layer as training progresses. AdaREPA uses a softmax distribution over smoothed layer compatibility scores to sample a layer for alignment at each training step. This probabilistic routing adapts teacher supervision to the student's changing learning needs, with minimal training overhead and no additional inference cost. On ImageNet-256, AdaREPA accelerates convergence and improves generation quality across model sizes, while revealing how the most suitable layers for supervision shift over training.

cs.CV

新论文 · v1arXiv:2611.24379

MoE-Context: Show the Model Only What It Needs

Vedant Navale

原文摘要

AI agents read their whole history at every step. People do not work like that: we keep what the moment needs and hold the rest as a rough note we can recall. MoE-Context applies this to language models. The context is split into blocks. Before each step a fast yes or no pass decides which blocks the step needs; the rest stay as one-line notes that can be brought back, and nothing is deleted. In the version we build, one model makes this decision and then answers, over a single key-value cache where hidden blocks are masked rather than removed. On a laptop with Qwen3-1.7B and a synthetic lookup task, showing only the needed block raises accuracy from 93 to 119 of 120, and a masked block is as hidden as a deleted one. The model's own picking is the weak part. A supervised linear probe written into spare output rows lifts its accuracy from 80 to 110 of 120, but simple keyword search picks perfectly on this task, and the probe does not transfer to a second task. We report these results openly, along with two cheap usage patterns that block switching enables: undoing a failed attempt and running subagents without a new context.

cs.AI · cs.CL · cs.LG