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

+479新增论文
+361其中 AI 相关
+64已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-16 01:34 · 全站最近入库:2026-11-16 21:56各信源进度 →

统计口径与数据新鲜度

统计 2026-11-16 03:10 — 2026-11-17 03:10(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:2026-11-16 01:34最近发现新增版本:2026-11-16 21:56库内论文最新更新时间:2026-11-14 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。

23,827 已收录论文19,941 AI 相关6,922 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 23,827 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-16 01:33(北京时间)
版本更新 · v2arXiv:2605.05745

Best Arm Identification in Generalized Linear Bandits via Hybrid Feedback

Qirun Zeng, Xuchuang Wang, Jiayi Shen, Xutong Liu, Fang Kong, Jinhang Zuo

原文摘要

We study fixed-confidence best arm identification in generalized linear bandits under a hybrid feedback model: at each round, the learner may query either (i) absolute reward feedback from a single arm or (ii) relative (dueling) feedback from an arm pair, both governed by generalized linear models. We introduce a likelihood-ratio--based confidence sequence that unifies heterogeneous generalized linear observations and yields an explicit ellipsoidal confidence set under a self-concordance assumption. Building on this confidence set, we propose a hybrid Track-and-Stop algorithm that adaptively allocates queries by tracking a minimax-optimal design over a joint action space of arms and pairs. We establish $δ$-correctness and provide almost-sure asymptotic upper bounds on the stopping time. We further extend the framework to a cost-aware setting that accounts for heterogeneous acquisition costs across feedback modalities. Empirical experiments demonstrate that the proposed algorithms significantly improve sample efficiency over baseline methods.

cs.AI

版本更新 · v3arXiv:2605.15960

Imperfect World Models are Exploitable

Logan Mondal Bhamidipaty, Esmeralda S. Whitammer, David Abel, Mykel J. Kochenderfer, Subramanian Ramamoorthy

原文摘要

We propose a novel definition of model exploitation in reinforcement learning. Informally, a world model is exploitable if it implies that one policy should be strictly preferred over another while the environment's true transition model implies the reverse. We analogize our definition with a prior characterization of reward hacking but show that the associated proof of inevitability does not transfer to exploitation. To overcome this obstruction, we develop a general theory of reward hacking and model exploitation that proves that exploitation is essentially unavoidable on large policy sets and yields the corresponding claim for hacking as a special case. Unfortunately, we also find that the conditions that guarantee unhackability in finite policy sets have no counterpart that precludes exploitation. Consequently, we introduce a relaxed notion of exploitation and derive a safe horizon within which it can be avoided. Taken together, our results establish a formal bridge between reward hacking and model exploitation and elucidate the limits of safe planning in world models.

cs.AI · cs.LG

版本更新 · v2arXiv:2602.21723

LessMimic: Versatile Humanoid-Object Interaction with Unified Distance Field Representations

Yutang Lin, Jieming Cui, Yixuan Li, Wei Liang, Baoxiong Jia, Yixin Zhu, Siyuan Huang

原文摘要

Humanoid robots capable of versatile whole-body interaction with everyday objects represent a central goal of embodied intelligence. Existing approaches often rely on motion-reference inputs or task-specific rewards, coupling policies to particular motion scripts, object geometries, and contact timings. This limits geometric generalization and flexible interaction control, where simple commands specify motion intent while object geometry determines how contact should be executed. We introduce LessMimic, a framework for versatile motion-free-at-inference humanoid-object interaction. At inference, a single whole-body policy is steered by root commands and an interaction-type flag instead of motion-reference inputs, and is conditioned on a compact interaction representation built from short histories of distance-field-derived surface distances and local surface directions. This geometry-conditioned interface enables command-steered control while adapting contact behavior to local object geometry. Through visual distillation, LessMimic further enables egocentric-depth-based deployment for MoCap-free sensing. Across object scales from 0.4x to 1.6x, LessMimic maintains robust performance over four interaction tasks, including PickUp, SitStand, Push, and Carry, where motion-conditioned baselines degrade sharply away from the training scale. Beyond single-task evaluation, LessMimic further supports sequential composition of heterogeneous interaction skills, attaining 62.1% success on five-task sequences and retaining 23.5% success at 15 task instances. By grounding interaction control in local geometry rather than motion references, LessMimic provides a path toward geometry-generalizable, command-steerable humanoid-object control with heterogeneous skill composition.

cs.RO

版本更新 · v7arXiv:2604.02923

An Inspectable LLM Council for Multi-Model Research Answer Aggregation

Shuai Wu, Xue Li, Zhijun Wang, Bolun Liu, Weilin Cai

原文摘要

Long-form research questions often require answers that combine complementary details, reconcile conflicting estimates, and retain qualifications across several large language model outputs. We present Standard Council, a two-stage synthesis workflow in which an analyst organizes three independent candidate answers into a structured textual state and a separate writer composes the final response from the question and that state. We evaluate the workflow on 100 closed-book tasks from the Deep Research Accuracy, Completeness, and Objectivity benchmark, DRACO, retaining 600 final answers, 100 analyst states, and 720 whole-answer judgments. Under the primary rubric judge, Council scores 70.98, exceeding GPT-5.6 Sol by 6.18 points and Gemini 3.8 Flash by 8.34 points; its 0.47-point difference from Claude Opus 5 has a 95% paired interval of $[-0.69,1.61]$. In the four-arm fusion comparison, Council scores 1.87 points above direct fusion, with a 95% paired interval of $[0.33,3.48]$. It covers 51.1% of positive criteria met by exactly one candidate, compared with 43.7% for direct fusion. On an exploratory 68-task subset with matching returned model labels, the Council-minus-direct-fusion difference is $-1.69$ points, with a 95% interval of $[-2.80,-0.62]$. The full comparison characterizes workflows that differ in structure and generation budget; estimating the analyst stage's effect requires a controlled component study. Traced cases show how selected estimates, provenance qualifications, and selection decisions pass through the saved state into final answers.

cs.CL · cs.AI

新论文 · v1arXiv:2611.10659

Pandemonium: Longform Target-Speaker Speech Recognition with Cross-Speaker Interactions

Alexander Polok, Dominik Klement, Ivan Medennikov, Antoine Laurent, Petr Pálka, Samuele Cornell, Shinji Watanabe, Sanjeev Khudanpur 等 12 位作者

原文摘要

Diarization-conditioned target-speaker ASR (TS-ASR), in particular Diarization-conditioned Whisper (DiCoW), is an accurate and data-efficient approach to multi-talker transcription. It has, however, three structural limitations: the model transcribes each speaker independently, operates on short (30 s) windows, and optimizes its diarization and TS-ASR components separately. We address the first two. We integrate the diarization conditioning into a longform Parakeet backbone (DiCoP) and add cross-speaker attention, which lets the model attend across the speaker dimension so that speakers are no longer decoded in isolation, and an STNO mask refiner that re-estimates the diarization-derived conditioning. As conversational data with genuine multi-speaker interaction is scarce, we scale training with a pseudo-labeling pipeline that mines 5.4k hours of in-the-wild multi-speaker audio from YODAS2. Combining these with synthetic and real data yields state-of-the-art results across several meeting benchmarks. We open-source the resulting models, DiCoP and Pandemonium, and the pseudo-labeled data.

eess.AS · cs.CL · cs.SD

版本更新 · v2arXiv:2503.07324

Decision-Dependent Stochastic Optimization: The Role of Distribution Dynamics

Zhiyu He, Saverio Bolognani, Florian Dörfler, Michael Muehlebach

原文摘要

Distribution shifts have long been regarded as troublesome external forces that a decision-maker should either counteract or conform to. An intriguing feedback phenomenon termed decision dependence arises when the deployed decision affects the environment and alters the data-generating distribution. In the realm of performative prediction, this is encoded by distribution maps parameterized by decisions due to strategic behaviors. In contrast, we formalize an endogenous distribution shift as a feedback process featuring nonlinear dynamics that couple the evolving distribution with the decision. Stochastic optimization in this dynamic regime provides fertile ground to examine the various roles played by dynamics in the composite problem structure. To this end, we develop an online decision-making algorithm that both adapts to and shapes the dynamic distribution. Throughout the paper, we adopt a distributional perspective and demonstrate how this view facilitates characterizations of distribution dynamics and the optimality and generalization performance of the proposed algorithm. We showcase the theoretical results in an opinion dynamics context, where an opportunistic party maximizes the affinity of a population with evolving, polarized opinions, and in a recommender system scenario, featuring performance optimization with discrete distributions in the probability simplex.

math.OC · cs.LG · eess.SY

版本更新 · v4arXiv:2505.21938

Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection

Qirun Zeng, Eric He, Richard Hoffmann, Xuchuang Wang, Jinhang Zuo

原文摘要

Adversarial attacks on stochastic bandits have traditionally relied on some unrealistic assumptions, such as per-round reward manipulation and unbounded perturbations, limiting their relevance to real-world systems. We propose a more practical threat model, Fake Data Injection, which reflects realistic adversarial constraints: the attacker can inject only a limited number of bounded fake feedback samples into the learner's history, simulating legitimate interactions. We design effective attack strategies under this model, explicitly addressing both magnitude constraints (on reward values) and temporal constraints (on when and how often data can be injected). Our theoretical analysis shows that these attacks can mislead a class of bandit algorithms into selecting a target arm in nearly all rounds while incurring only sublinear attack cost. Experiments on synthetic and real-world datasets validate the effectiveness of our strategies, revealing vulnerabilities in stochastic bandit algorithms under practical adversarial scenarios.

cs.LG · cs.AI · cs.CR

新论文 · v1arXiv:2611.10658

The Delta Gap: Cellular World Models for Perturbation Response Prediction Are Far Harder Than Global Correlation Implies

Junhui Gao

原文摘要

Cellular world models aim to predict how cells respond to genetic perturbations, analogous to world models in reinforcement learning. Perturbation response models are evaluated almost exclusively by global Pearson correlation between predicted and observed transcriptomes. We show this metric is dominated by thousands of non-responsive genes and severely inflates apparent performance. On the Replogle K562 screen (509 unseen perturbations, 5000 genes), GEARS achieves global Pearson 0.977, but its delta Pearson is only 0.076. A trivial baseline predicting mean training delta achieves 0.267, which is 3.5x better than GEARS. Our primary contribution is methodological: we identify and quantify this delta gap as a systematic evaluation artifact, and propose delta-space evaluation as the primary benchmark. We provide two post-hoc corrections (RWR on co-expression graphs and kNN matching) and show that 2.8x of the remaining gap lies in perturbation representation geometry, not decoder capacity. We further show the gap is partially addressable end-to-end: delta-aligned training (lambda = 3.0) raises delta Pearson to 0.179 on Replogle (2.35x over GEARS) and surpasses the trivial baseline on direction accuracy (0.680 vs 0.567) and gene recovery (1.52 vs 1.19); combined with kNN it reaches 0.217. An amplitude-calibration variant (eta = 0.3) further raises gene recovery to 2.16 while reaching 0.218. Critically, these representations transfer zero-shot across cell lines: from Replogle K562 to unseen Norman A549, the delta-aligned model reaches 0.328 on 741 shared genes, 4.7x a trivial baseline and 2.1x standard GEARS. This is the first systematic quantification of the delta gap, and the first diagnostic framework localizing the residual gap to perturbation representation geometry via an oracle upper bound.

cs.AI · cs.LG

版本更新 · v2arXiv:2611.04483

CogniLife: Continuous Cognitive Distillation for Deployed Multi-Expert Agent Marketplaces

Xuwei Tan, Jingchun Huang, Ming Cai, Shuangquan Lyu, Tiange Wang, Yinzhi Wang, Ronghua Shi, Jingwei Song 等 15 位作者

原文摘要

Multi-expert agent marketplaces -- where creators publish domain-specific AI experts for end users to consult -- are an emerging deployment pattern for language technology. A published expert can receive many requests yet little evidence about its creator, who remains active elsewhere: the cold-expert problem. We frame continuous cognitive distillation as converting attributed creator observations into revisable decision rules, communication patterns, and domain procedures. CogniLife proposes a cross-expert attribution design for this problem, with Creator Profiles for transferable creator patterns and Expert Journals for listing-specific knowledge; this design is not yet implemented or evaluated. A deployed platform partially realizes the surrounding lifecycle through creator-authorized authoring, capability-owned proposal filtering, and reversible consolidation of eligible owner-scoped memory. On ProfyBench-v1, an 85-task offline benchmark built from replay seeds and controlled internal tests, the Profy Full configuration as a whole achieves a mean overall score of 4.65/5 from the primary LLM judge, compared with 4.01 for Raw LLM and 4.28 for Expert Prompt, with the largest gains in completeness, specificity, and missing-information tasks. These configuration-level scores do not test cold-expert mitigation, cross-expert transfer, or longitudinal distillation.

cs.AI

版本更新 · v2arXiv:2610.21361

Hierarchical Building-Floo-Room Framework for Privacy-Preserving Autonomous Delivery in Large-Scale Buildings

Xu Xiaoji, Chen Jinhua, Zhang Leihan, Luo Rui, Liao Zhongyuan

原文摘要

Autonomous delivery in large scale buildings requires coordinated planning across multiple spatial levels, including building, floor, and room, while ensuring scheduling efficiency. Existing methods often focus solely on algorithmic development, neglecting integration with physical robotic platforms, and rarely validate performance in real world scenarios. Many approaches also overlook hierarchical building structures and room level privacy constraints, limiting practical applicability. To address these challenges, this paper presents a hierarchical delivery framework based on a Building-Floor-Room abstraction, which decomposes the problem across multiple levels to significantly reduce optimization complexity. We formulate a building aware extension of the capacitated vehicle routing problem (CVRP) that explicitly models inter building and inter floor connectivity, room level access constraints, and privacy sensitive areas, enabling globally feasible and privacy compliant task assignment and routing. Based on this hierarchical model, the framework can evaluate delivery efficiency across different building layouts and optimize depot placement for specific configurations. At the execution level, we introduce a novel hardware design for non interactive, privacy preserving delivery, supported by reachability aware distances and a closed loop mapping, localization, and navigation system, ensuring consistency between planned routes and physical navigation. The proposed framework is validated on a real multi sensor robotic platform, demonstrating reliable multi floor delivery with improved path rationality and overall efficiency.

cs.RO

版本更新 · v3arXiv:2603.05093

From Baselines to Transport Geodesics: Axiomatic Attribution via Optimal Generative Flows

Cenwei Zhang, Lin Zhu, Manxi Lin, Lei You

原文摘要

Feature attributions often hide a critical modeling choice: they explain a prediction along a counterfactual path from a reference state to an input. Different baselines, interpolations, and generative trajectories define different paths and can therefore produce different explanations. We study this path ambiguity as a modeling problem. Our central question is whether the path can be chosen by a distribution-level transport process, rather than by a hand-designed interpolation or by the sensitivity geometry of the model being explained. We separate attribution into fixed-path credit allocation and path selection. For a fixed path, we prove that the Aumann-Shapley line integral is the unique attribution rule under fixed-path axioms and coordinate-trace separability. For path selection, we minimize kinetic action over flows that transport a reference distribution to the data distribution, yielding a transport-geodesic attribution principle. We instantiate this framework using Rectified Flow and derive stability bounds quantifying how vector-field approximation errors affect the resulting attributions. Experiments show that Reflow improves cross-seed attribution stability, while learned transport paths yield spatially coherent explanations with competitive deletion faithfulness among the evaluated path-based methods. Our code is available at https://github.com/cenweizhang/OTFlowSHAP.

cs.LG · cs.AI · cs.CV

版本更新 · v2arXiv:2602.06953

DAWN: Dependency-Aware Fast Inference for Diffusion LLMs

Lizhuo Luo, Zhuoran Shi, Jiajun Luo, Zhi Wang, Shen Ren, Wenya Wang, Tianwei Zhang

原文摘要

Diffusion large language models (dLLMs) have shown advantages in text generation, particularly due to their inherent ability for parallel decoding. However, constrained by the quality--speed trade-off, existing inference solutions adopt conservative parallel strategies, leaving substantial efficiency potential underexplored. A core challenge is that parallel decoding assumes each position can be filled independently, but tokens are often semantically coupled. Thus, the correct choice at one position constrains valid choices at others. Without modeling these positional dependencies, parallel strategies produce deteriorated outputs. Motivated by this insight, we propose DAWN, a training-free, dependency-aware decoding method for fast dLLM inference. DAWN extracts dependency relations and leverages two key observations: (1) masked positions dependent on unmasked certain positions become more reliable; (2) coupled masked positions strongly influence each other's predictions. Given those findings, DAWN leverages a dependency graph to select more reliable unmasking positions at each iteration, achieving high parallelism with negligible loss in generation quality. Extensive experiments across multiple models and datasets demonstrate that DAWN speedups the inference by 1.80-8.06x over baselines while preserving the generation quality. Code is released at https://github.com/lizhuo-luo/DAWN.

cs.CL

新论文 · v1arXiv:2611.10657

MMGraph-Agent: Agentic Multimodal RAG via Cache-Inspired Multimodal Knowledge HyperGraphs

Haoran Luo, Ziyue Zhu, Shangyang Wu, Linhao Luo, Yikai Guo, Qika Lin, Fangzhi Xu, Jiapu Wang 等 11 位作者

原文摘要

Driven by increasingly complex real-world applications, Retrieval-Augmented Generation (RAG) has evolved from textual pipelines to multimodal settings, comprising two complementary dimensions: knowledge organization and retrieval enhancement. However, existing methods still face challenges of high construction cost, context window limitations, and the lack of unified training across offline and online knowledge sources. To address these challenges, we propose MMGraph-Agent, a unified multimodal RAG framework based on cache-inspired multimodal knowledge hypergraphs. MMGraph-Agent enables extremely lightweight hypergraph construction with zero API cost and dynamic memory-based retrieval that alleviates long-context issues. Experiments across offline, online, and joint retrieval settings demonstrate consistent improvements in performance, efficiency, and architectural unification. Our software and data are publicly available at https://github.com/LHRLAB/MMGraph-Agent.

cs.AI

版本更新 · v2arXiv:2604.03556

Focus Matters: Attention-Value Dynamics for Efficient Hallucination Mitigation in Vision-Language Models

Sohyeon Kim, Sang Yeon Yoon, Kyeongbo Kong

原文摘要

Large Vision-Language Models (LVLMs) have achieved impressive progress in multimodal reasoning, yet they remain prone to object hallucinations, generating descriptions of objects that are not present in the input image. In this work, we investigate hallucination from the perspective of attention-value dynamics inside LVLM vision encoders. We identify a consistent three-phase structure of visual processing---diffusion, focus, and rediffusion---and show that the focus phase is where attention most clearly separates strongly and weakly supported visual tokens. However, low attention does not necessarily imply negligible downstream influence: low-attention tokens in the focus phase can still exert non-negligible value-side influence on the attention output relative to their small attention mass. Through controlled phase-wise value interventions, we find that hallucination behavior is particularly sensitive to the value content of low-attention tokens during this phase. Replacing or neutralizing these values reduces hallucination metrics while largely preserving grounded object evidence. A token-level teacher-forcing analysis further shows that the intervention reduces the probability of hallucinated object tokens with much smaller effects on ground-truth object tokens. In addition, Visual Attention Ratio (VAR) analysis shows that focus-phase intervention is accompanied by increased attention to visual tokens during decoding. Based on these observations, we instantiate a simple training-free inference-time intervention that replaces focus-phase low-attention values with an image-level mean value vector using statistics from a single forward pass. Experiments across multiple LVLM backbones demonstrate that this analysis-derived intervention reduces object hallucination with negligible additional runtime and remains compatible with existing inference-time mitigation methods.

cs.CV · cs.AI · cs.CL

新论文 · v1arXiv:2611.10653

FG-DSA: Frequency-Guided Depthwise Separable Attention for Efficient Dense Prediction

Yiyao Yang

原文摘要

Dense prediction and generative modeling in resource-constrained scenarios remain challenging, as repeated downsampling substantially reduces spatial detail and degrades boundaries. Although recent attention mechanisms have alleviated the problem of semantic ambiguity, they often fail to preserve high-frequency features or do so at the expense of a very high computational cost. In this paper, we propose the Frequency Guided Depthwise Separable Attention for Efficient Dense Prediction (FG-DSA), a novel framework for efficient dense visual understanding and generation. FG-DSA employs a 2D FFT high-pass filtering module to explicitly inject frequency domain priors into the spatial attention mechanism, allowing lightweight networks to recover sharp boundary transitions with linear complexity. A range of experiments show that FG-DSA is effective at recovering fine object boundaries and leads to a consistent performance gain in the task of semantic segmentation. Importantly, we show that FGA-DSA can be used as a module in diffusion-based generative U-Nets, leading to a significant improvement in the detail of the generated images with very little additional parameter and Floating Point Operation (FLOP) cost.

cs.CV

新论文 · v1arXiv:2611.10652

SHIN: Scalable Hierarchy for INcremental 3D Gaussian Splatting

Reagan Koo, Yeong-Gyu Kim, Eun-Seok Ryu

原文摘要

3D Gaussian Splatting (3DGS) provides high-quality real-time novel-view synthesis, but deployment budgets vary across devices and network conditions. Most compact methods target one operating point, so another budget requires a separate representation. SHIN is a scalable 3DGS hierarchy with a reusable Base and additive enhancement layers, so each higher-quality operating point extends rather than replaces the decoded representation. Constructing this hierarchy requires deciding which Gaussians each stage adds and how much spherical-harmonic (SH) appearance each retains, because these choices remain fixed in later prefixes. SHIN derives one shared Gaussian order from multi-scale position-scale sensitivity and divides each stage budget between adding Gaussians and retaining SH bands using calibrated gradient-based scores per byte. Lightweight opacity updates adjust compositing while transmitted geometry and SH coefficients remain unchanged. Across three benchmarks, SHIN progressively improves one cumulative representation and achieves the best PSNR-size efficiency in every overlapping comparison with the multi-rate and progressive baselines.

cs.CV

版本更新 · v2arXiv:2606.31963

Signed-Permutation Coordinate Transport for RMSNorm Transformers

John Sweeney

原文摘要

Modern LLM workflows move coordinate-indexed objects across checkpoints: steering vectors, sparse autoencoders, top-$k$ neuron sets, attribution lists, LoRA updates, optimizer states, and merge alignments. This requires fixing the model's residual-stream gauge. We show that the native coordinate-preserving orthogonal gauge is architecture-dependent: LayerNorm residual charts have permutation gauge $S_d$ up to a global sign flip, while RMSNorm charts with generic gains have signed-permutation gauge $B_d = S_d \ltimes \{\pm 1\}^d$. Permutation-only alignment is thus symmetry-incomplete for RMSNorm models. We introduce sign-marginalized Hungarian matching and prove that with decorrelated coordinates, signed-correlation matching has a permutation-accuracy ceiling equal to the true gauge's positive-sign fraction up to $O(d 2^{-d})$, which sign-marginalized matching removes. Omitting signs induces expected squared error $2\|ΔW\|_F^2$ on weight updates and loses orientation on vectors and AdamW first moments, while RMSNorm gains and AdamW second moments permute without signs. Controlled tests confirm these laws: a Qwen2.5-1.5B SST-2 LoRA adapter retains 0.944 accuracy (from 0.945) under $B_d$ transport vs. 0.523 under $S_d$, and steering preserves effects to $<10^{-4}$ at zero raw top-10 overlap. Across natural checkpoints, inherited slots suffice for same-base descendants (96.2%-100% retention), whereas cross-generation pairs support sparse prompt-stable subsets with causally decisive negative signs ($+0.185$ vs. $-0.010$ at $8σ$ on Qwen2/2.5-0.5B; $+0.189$ vs. $-0.021$ on 1.5B). Along fine-tuning trajectories, composing local $B_d$ gauges recovers 91.1% of cross-run Qwen2.5-1.5B coordinates at 1,500 steps vs. 60.3% for endpoint matching. Merge and interpretability audits confirm that coordinate names are reproducible only relative to an explicit $B_d$ gauge.

cs.LG · cs.CL · stat.ML

版本更新 · v3arXiv:2610.17623

RELAY: Selective Self-Distillation with Adaptive Hand-off for Rubric-Based Reinforcement Learning

Kunwang Zheng, Xinyu Sun, Yu Cao, Shuocheng Li, Lei Cheng, Ying Liu, Qilong Zhang, Jinlong Chen 等 10 位作者

原文摘要

Rubric-based reinforcement learning grades a response against many written criteria but collapses them into one reward, which discards the detail the rubric carries and gives no GRPO reward gradient when a group's responses tie. A frozen teacher that reads the rubric as privileged context provides token-level supervision, but we observe an early plateau in the gains over GRPO. RELAY uses advantages and rewards to select positive-advantage and low-scoring zero-advantage rollouts, and monitors the distillation loss to decide when to retire the teacher. Stage 1 combines selective distillation from a frozen teacher with GRPO; a plateau detector then removes the teacher, and Stage 2 continues with GRPO alone. Calibrated on one model-domain pair and reused unchanged in three others, the plateau detector fires between steps 122 and 171. RELAY scores highest in-domain among the compared methods in all four settings, ahead of GRPO on RaR-Medicine by 4.62 points with Qwen3-1.7B and 6.02 with Qwen3-4B. On Qwen3-1.7B/Medicine, removing the teacher from a shared checkpoint outperforms keeping, weakening or refreshing it. Distilling selected rollouts covering about half the response tokens also outperforms distilling all rollouts at a matched average coefficient.

cs.LG · cs.AI · cs.CL

版本更新 · v3arXiv:2602.20031

Latent Introspection: Models Can Detect Prior Concept Injections

Theia Pearson-Vogel, Martin Vanek, Raymond Douglas, Jan Kulveit

原文摘要

We uncover a latent capacity for introspection in a Qwen 32B model, demonstrating that the model can detect when concepts have been injected into its earlier context and identify which concept was injected. While the model denies injection in sampled outputs, logit lens analysis reveals clear detection signals in the residual stream, which are attenuated in the final layers. Furthermore, prompting the model with accurate information about AI introspection mechanisms can dramatically strengthen this effect: the sensitivity to injection increases massively (0.3% -> 39.9%) with only a 0.6% increase in false positives. Also, mutual information between nine injected and recovered concepts rises from 0.61 bits to 1.05 bits, ruling out generic noise explanations. Our results demonstrate models can have a surprising capacity for introspection and steering awareness that is easy to overlook, with consequences for latent reasoning and safety.

cs.AI · cs.LG

新论文 · v1arXiv:2611.10650

Trajectory Inversion for a Rudderless Dual Vectored-Thrust Flying Wing

Weicheng Di, Qiqi Liu, Jinwu Xiang, Daochun Li, Zhan Tu

原文摘要

A rudderless dual vectored-thrust flying wing combines vertical takeoff and landing with efficient for- ward flight in a compact tailless configuration. Without aerodynamic control surfaces, vectored thrust couples translation and pitch control. Therefore, generating dy- namically compatible trajectories remains an open chal- lenge. To address this challenge, this study proposes a trajectory generation method that jointly optimizes path geometry and timing. Trajectory inversion yields state and control histories that satisfy the coupled equations of motion. The optimization balances curvature-induced acceleration against aerodynamic support, lowering thrust- vectoring demand while retaining the defining maneuver features. Representative looping and rolling sequences are tracked in closed-loop simulations. These results provide a basis for planning autonomous aerobatic maneuvers with rudderless vectored-thrust aircraft.

cs.RO

版本更新 · v3arXiv:2507.15778

Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR

Runze Liu, Jiakang Wang, Fuzheng Zhang, Xiu Li, Guorui Zhou, Kun Gai, Ling Pan

原文摘要

Reinforcement Learning with Verifiable Rewards (RLVR) has become an effective post-training method for improving the reasoning abilities of Large Language Models (LLMs). However, existing methods mainly apply uniform optimization constraints across all tokens, ignoring their heterogeneous roles. Prior work shows that high-entropy tokens are closely tied to reasoning, while low-entropy tokens primarily encode factual knowledge, and recent approaches attempt to exploit this distinction by isolating token updates via masking or asynchronous training. We argue that retaining training signals across token types can lead to more effective learning. To address this, we propose Archer, an entropy-aware RLVR framework with dual-token constraints that preserves joint optimization while modulating update strength across token types. Our method introduces response-level entropy normalization for stable token classification and applies differentiated clipping ranges and KL regularization to encourage exploration on reasoning tokens while preserving knowledge tokens. Experiments on mathematical reasoning and code generation benchmarks show that Archer consistently outperforms strong baselines across multiple model scales, improving both pass@1 and pass@K performance. These results highlight the importance of respecting sequence-level dependencies when designing fine-grained RL optimization strategies for LLMs. The code is available at https://github.com/wizard-III/ArcherCodeR.

cs.CL

版本更新 · v2arXiv:2605.10206

Extended Wasserstein-GAN Approach to Causal Distribution Learning: Density-Free Estimation and Minimax Optimality

Shu Tamano, Masaaki Imaizumi

原文摘要

Distributional causal inference requires estimating not only average treatment effects but also interventional outcome distributions, including quantiles, tail risks, and policy-dependent uncertainty. As a method for distributional causal inference, generative adversarial network (GAN)-based counterfactual methods are flexible tools for this task. However, these methods have several limitations. First, the objectives of certain techniques do not coincide with the statistical risk of the identifiable causal target, and therefore provide limited theoretical guarantees regarding estimable counterfactual distributions or optimality. Second, implementing target-design risks under observational sampling can require density-ratio estimation. In this paper, we propose GANICE (GAN for Interventional Conditional Estimation) with several advantages: it (i) specifies the conditional interventional distribution for each treatment--covariate state as the causal estimation target; (ii) estimates the conditional distribution such that its averaged Wasserstein risk is minimized; (iii) attains minimax optimality. GANICE achieves these advantages through the extended Wasserstein distance and cell-normalized estimation, with minimax bounds based on Wasserstein GAN oracle inequalities and an anisotropic Assouad construction; Besov theory characterizes the cellwise critics. Our experiments show that GANICE outperforms existing methods on most benchmarks.

math.ST · cs.LG · stat.ML

版本更新 · v2arXiv:2602.07603

Closed-Form Implicit Neural Representations

Woojin Cho, Junghwan Park

原文摘要

Training implicit neural representations (INRs) to capture fine-scale details typically relies on iterative backpropagation and is often hindered by spectral bias when the target exhibits highly non-uniform frequency content. We propose ELM-INR, a backpropagation-free INR that decomposes the domain into overlapping subdomains and fits each local problem using an Extreme Learning Machine (ELM) in closed form, replacing iterative optimization with stable linear least-squares solutions. This design yields fast and numerically robust reconstruction by combining local predictors through a partition of unity. To understand where approximation becomes difficult under fixed local capacity, we analyze the method from a spectral Barron norm perspective, which reveals that global reconstruction error is dominated by regions with high spectral complexity. Building on this insight, we introduce BEAM, an adaptive mesh refinement strategy that balances spectral complexity across subdomains to improve reconstruction quality in capacity-constrained regimes.

cs.LG · math.NA

新论文 · v1arXiv:2611.10647

Which Taxels, Not How Many: Tactile Sensor Placement for Dexterous Policies

Taeyun Kim, Jinsoo Shin, Taehwa Kim, Taeyoon Kim, Jinyoung Choi

原文摘要

Policies for dexterous manipulation depend on design choices made before training: which sensors enter the observation, where they are placed, and how their readings are arranged for the encoder. These choices are rarely examined. For tactile sensing, simulation studies place taxels uniformly over the hand, and this placement has become a default without being tested. We test this default under a single controlled protocol. The protocol holds the teacher, the training procedure, and the candidate taxels fixed and varies only the sensor configuration and the representation. We evaluate it on in-palm rotation and grasping with three robot hands. A small set of taxels selected from contact statistics matches a far larger uniform set. Taxels at rarely contacted sites are equivalent to no tactile input. With all candidate taxels, their layout in the input tensor does not affect performance, even for a convolutional encoder. Uniform placement therefore overestimates the number of taxels a task requires. We distill these results into lessons for designing tactile policies and into a minimal set of design choices that a tactile policy study should report.

cs.RO

版本更新 · v2arXiv:2606.26631

Adaptive Visual KV Cache Reuse for Efficient Multi-Turn Iterative Multimodal Reasoning

Mengzhao Wang, Yanli Ji, Wangmeng Zuo, Chongjun Tu

原文摘要

Multi-turn visual reasoning acquires evidence through repeated cropping and local inspection, but overlapping observations lead to redundant visual encoding and prefill. We propose an adaptive visual KV cache reuse framework that reduces repeated processing while retaining access to additional image detail. The central difficulty is that skipping visual prefill also removes the interaction between a new observation and the current reasoning: historical keys remain bound to old positions, while historical values have not incorporated the intent formed between crops. Our framework first selects full re-encoding or regional reuse according to crop size and cache availability. On the reuse path, Positional Key Rebinding (PKR) corrects attention matching at the new positions, and Intent-guided Value Enhancement (IVE) uses pre-crop reasoning attention to strengthen content relevant to the current objective. Source states from fully re-encoded crops are retained for subsequent retrieval and reuse. The framework requires no additional training on reuse hits, it bypasses the visual encoder and prefill for the selected visual tokens, then resumes normal text reasoning. Experiments show that our method reduces inference time while preserving the overall accuracy of multi-turn visual reasoning that relies on intermediate crop images.

cs.CV

新论文 · v1arXiv:2611.10646

On the Geometric Bottlenecks of Cross-Lingual Knowledge Editing

Aastha A K Verma, Anwoy Chatterjee, Sparsh Jain, Tanmoy Chakraborty

原文摘要

Transformer-based autoregressive multilingual language models (LMs) are often assumed to store factual knowledge in an interlingua: a language-agnostic latent space. If this assumption holds, a parameter update applied in a source language during knowledge editing should inherently propagate to its semantic equivalents in target languages. However, empirical benchmarks show that cross-lingual propagation of edits often degrades substantially, a degradation that prior work treats largely as an algorithmic symptom rather than a structural bottleneck. In this work, we investigate whether representational anisotropy creates a structural bottleneck for source-only associative-memory editing, and perform a geometric dissection of the cross-lingual propagation gap. Using associative memory-based "locate-then-edit" methods, including ROME, MEMIT, and AlphaEdit, as analytical frameworks, we derive a closed-form decomposition of this gap into key-space propagation, parallel residual mismatch, and orthogonal residual mismatch. Our empirical findings show that cross-lingual propagation is primarily bottlenecked by a large orthogonal residual component, indicating that source-language edits often fail to span the target-language update directions required for successful transfer. Furthermore, we show that post-hoc correction using only source-language residual statistics encounters a large residual-prediction barrier. Finally, we perform a structural realignment via a targeted contrastive alignment objective and find that it substantially improves key-space propagation and partially improves cross-lingual edit transfer, while leaving value-space mismatch as a remaining bottleneck.

cs.CL · cs.LG

新论文 · v1arXiv:2611.10644

Agent2BIM: A Systematic Evaluation of Reality-Capture Configurations and Prior Models for As-Built Modeling

Boyu Wang, Borja García de Soto

原文摘要

AI agents can generate as-built Building Information Models (BIMs) from reality-capture data, but the evidence required for accurate reconstruction remains unclear. This paper presents Agent2BIM, an agent-based Scan-to-BIM framework, and evaluates seven capture configurations in a research laboratory and a mechanical, electrical, and plumbing (MEP) scene. Inputs include 360° panoramas, smartphone video, smartphone LiDAR, terrestrial laser scanning (TLS), TLS with posed panoramas, and imagery supplemented by prior IFC models. Outputs are compared against TLS-verified reference models for component detection, geometric accuracy, and MEP connectivity. In the laboratory, component recall ranged from 84% to 100%, but only 9% of components reconstructed from panoramas were within 20 cm of their reference positions, compared with 86% using smartphone LiDAR. In the MEP scene, scale-free inputs recovered 11 - 43% of components, versus 75 - 93% for metric inputs. Prior IFC models improved MEP reconstruction but could propagate inaccuracies. The results highlight the complementary roles of visual and metric evidence in agent-based Scan-to-BIM.

cs.CV

新论文 · v1arXiv:2611.10642

VisProve: Learning Typed Visual Provenance for Scientific Document Question Answering

Jingxuan Wei, Zhengyang Wang, Zheng Guo, Xiaoming Wu, Fuqiang Wang, Xin Wang, Bihui Yu, Yining Wang 等 9 位作者

原文摘要

Visual citations show where an answer is supported, but not how the cited content contributes to it. In scientific document question answering, a model may read a value directly, compress several observations, or combine evidence to infer a conclusion. We study typed visual provenance, where a system generates an answer together with source regions and explicit support relations. We introduce ReViE, a dataset of 3,941 questions over complete scientific documents, with page, section, and region annotations labeled as Quotation, Compression, or Inference. We further propose VisProve, which augments joint answer and provenance generation with differentiable Spatial Evidence Reading and Evidence-Answer Relation Feedback. VisProve conditions next-token prediction on localized visual evidence and its estimated support relation through a gated norm-bounded residual. On the document-disjoint ReViE test set, VisProve improves Typed-F1 from 3.52 to 5.48 and Section-F1 from 45.99 to 53.24 over joint supervised fine-tuning with the same backbone, while Page-F1 changes only marginally. These results suggest that coupling visual grounding with support relation modeling mainly improves fine-grained provenance, while the low absolute Typed-F1 reveals a substantial gap between answer generation and precise visual evidence attribution.

cs.CV · cs.AI

版本更新 · v4arXiv:2601.22197

Neural Signals Generate Clinical Notes in the Wild

Jathurshan Pradeepkumar, Zheng Chen, Jimeng Sun

原文摘要

Generating clinical reports that summarize abnormal patterns, diagnostic findings, and clinical interpretations from long-term EEG recordings remains labor-intensive. We present CELM, the first clinical EEG-to-Language foundation model capable of summarizing long-duration, variable-length EEG recordings and performing end-to-end clinical report generation at multiple scales. CELM integrates pretrained EEG foundation models with language models to enable scalable multimodal learning. We curate a large-scale clinical EEG dataset containing 9,922 reports paired with approximately 11,000 hours of EEG recordings from 9,048 patients to train CELM, and release the benchmark pipeline, with automated report-structuring, to facilitate future research. Experimental results show that CELM consistently outperforms existing methods across all evaluation settings. Importantly, we further conduct human evaluation with clinical experts, demonstrating that CELM generates reports that are more clinically coherent, diagnostically reliable, and better aligned with expert interpretation. We release our model code and benchmark construction pipeline at https://github.com/Jathurshan0330/CELM.

cs.LG · cs.AI · eess.SP

新论文 · v1arXiv:2611.10641

Same Contacts, Different Rooms:Meeting Format Shapes LLM Group Influence Beyond Pairwise Contact Counts

Zhexi Gu, Shusheng Cao, Runping Zhu

原文摘要

Simulations of LLM agent collectives usually summarise interaction as a pairwise contact network. We test that summary with a contact-matched intervention: the members, their initial stances and every pairwise contact count stay fixed while the meetings are reorganised, and each member gives a one-step support readout after reading the arguments of everyone present. Across six open-weight LLMs and 14 debate topics, merging triads into five-person meetings (resize) moves group support more than reshuffling who meets in triads (regroup), although both move the same 60 dyadic exposures. The root-mean-square change is 4.36 versus 2.59 percentage points (pp), larger for resize in all six models and in 81 of 84 model-topic cells. Resizing also has a consistent direction: it amplifies the initial majority of the merged members by 1.72 pp, in every model, whereas regrouping has no direction (<= 0.03 pp). A contact network cannot represent regrouping at all, yet pairwise rules applied meeting by meeting explain up to 76% of its effect, and with identical weights they raise resize R^2 by 0.51-0.58 for two models. We show that the gap between the two placements is an exact Jensen gap of the sigmoid readout, bounded by how much a member's predicted logits vary across meetings; this dispersion tracks the gap with log-log r=0.97 and tells a modeller when contact counts are a safe summary. Meeting format is a first-class variable for LLM social simulation.

physics.soc-ph · cs.CL