跳到正文
研踪PaperTrail

YOUR RESEARCH, IN SIGHT

让下一篇重要论文，不再错过。

覆盖计算机全学科，以 AI 为快捷入口。关注方向、作者和关键词，把发现变成持续的研究积累。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+0新增论文
+0其中 AI 相关
+1已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-08 04:03 · 全站最近入库：2026-11-07 17:43各信源进度 →

统计口径与数据新鲜度

统计 2026-11-07 04:22 — 2026-11-08 04:22（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：2026-11-08 04:03最近发现新增版本：2026-11-07 17:43库内论文最新更新时间：2026-11-06 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。

17,100 已收录论文14,661 AI 相关4,056 有修订112 HF 社区发现

浏览论文库

研究日报 / 周报 →按信源浏览 →

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 17,100 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-08 04:02（北京时间）
新论文 · v1arXiv:2611.04806

On the Complexity of Finite-Sum Nonconvex-Strongly-Concave Minimax Optimization with Exact Proximal Oracles

Yuxing Peng, Zhiqing Tang, Weijia Jia

原文摘要

Finite-sum nonconvex-strongly-concave minimax problems arise in adversarial training and distributionally robust learning. Their known lower bounds cover only linear-span algorithms, zero-respecting algorithms, or a restricted proximal protocol. They leave open the power of exact component proximal steps, whose responses are global saddle points of regularized component problems and can expose directions that gradients do not reveal. We show that neither unrestricted adaptivity nor exact proximal access improves the polynomial complexity. We consider $n$ components with joint mean-squared smoothness $L$ and condition number $κ$, started from a point with primal gap at most $Δ$ and joint gradient norm at most $\sqrt{2LΔ}$. For $\varepsilon^2/(LΔ)$ below a universal constant, every randomized adaptive algorithm with exact component values, gradients, and proximal responses needs $Ω(n+\min\{\sqrt n\,κ,n^{3/4}\sqrtκ\}LΔ/\varepsilon^2)$ component queries in the worst case to find an $\varepsilon$-stationary point of the value function. A method using only component gradients attains this rate up to a factor $\log^2(2+κ)$, including initialization. Zhou (arXiv:2610.04944) concurrently proves the same rate for zero-respecting first-order methods. In our hard instances, every component admits proximal steps with the same step sizes, and a response for a component reveals hidden directions only up to the first one assigned to another component.

cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2607.13284

Tight Bounds for $(1+ε)$-Approximation for Minimum Spanning Trees in Doubling Metrics

Jan Höckendorff, Felix Hommelsheim, Christian Sohler, Di Yue

原文摘要

The minimum spanning tree (MST) problem is one of the most basic optimization problems on metric spaces and graphs. We study the problem of computing a $(1+ε)$-approximation to the MST of an $n$-point metric space of doubling dimension $\mathrm{ddim}$. In doubling metrics, previous deterministic algorithms incur a running time with dependence $ε^{-O(\mathrm{ddim})}$. We give a deterministic algorithm that computes a $(1+ε)$-approximate MST in time $$ O_{\mathrm{ddim}}\left(n\log n+ \min\left\{n^2,\frac{n}ε\log(1/ε)\right\}\right) $$ in the standard real RAM model with a unit-cost exact-distance oracle. For bounded doubling dimension, this improves the previous dependence on $ε$ to essentially linear in $ε^{-1}$. We complement this algorithm with a lower bound of $$ Ω\!\left(n\log n+ \min\left\{n^2,\frac{n}ε\log(1/ε)\right\}\right) $$ on running time in the same model, already for metrics of doubling dimension at most seven, showing that our running time is essentially tight. For points in $\mathbb R^d$ with $d = O(1)$, exploiting Euclidean geometry allows us to further improve the dependence on $ε$. We obtain a $(1+ε)$-approximate MST in expected time $$ O_d\left( n \log n + \min\left\{n^2, \frac{n}{ε^{1 - β_d}} \log^{1-β_d}(1/ε) \right\} \right) $$ for some $β_d=Θ(1/d)$, improving the accuracy-dependent term $nε^{-2}\log^2(1/ε)$ of Arya and Mount [SODA'16].

cs.DS

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04805

Where Is the Truth When AI Agents Disagree? Dependence-Aware Belief Propagation for Unsupervised Truth Discovery (Extended Version)

Laure Berti-Equille

原文摘要

LLM agents often give conflicting answers to the same question, and many deployed agents are variants of one base model, so their errors are correlated. We study truth discovery among such agents: inferring the true answer and each agent's reliability without labels. Classical truth-discovery and crowdsourcing estimators reward an agreeing bloc and can fall below majority vote. We propose BP-TD, which adds per-agent decorrelation weights to mean-field variational inference on a one-coin Dawid-Skene model, so that agents sharing a base model count together as about one vote. Its first iteration is exactly a weighted vote; later iterations estimate agent reliability. With 13 local agents on MMLU, BP-TD beats majority vote without any labels (0.716 vs. 0.624 on 160 questions; 0.707 vs. 0.674 on 10,000) and ranks agents by their true accuracy, which classical estimators invert. The gain depends on the pool: over 2,852 sub-pools of four benchmarks, decorrelation helps when the largest correlated bloc is weaker than the other agents and hurts when it is the strongest, and we found no label-free statistic that tells the two cases apart. Labels change the picture quickly: with 20 labelled questions, selecting the single best agent is more accurate than every aggregator we tested on MMLU, ARC and TruthfulQA.

cs.AI · cs.CL · cs.DB · stat.ML

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.42147

Latent Sparse Attention: Towards Subquadratic Ultra-Long Context Attention

Ales Kubicek, Andrea Caciolai, Liangzu Peng, Sukrit Sumant, Benjamin Hoover, Fraser Greenlee, Ruairidh M. Battleday, Radim Urban 等 10 位作者

原文摘要

Agentic workflows are increasingly shifting towards long-horizon tasks, such as open-ended scientific work, where models benefit from long context windows. However, at ultra-long context (above 1M tokens), full attention is prohibitively expensive. A promising approach is sparse attention, in which a dedicated indexer selects a subset of keys for the attention. Yet current methods of indexing either still have quadratic cost in context size or lose accuracy at ultra-long context scale. We address this with Latent Sparse Attention (LSA), an efficient attention method that maintains accuracy at ultra-long context. In LSA, keys are hierarchically clustered in a learned latent space, and queries score the hierarchy of clusters instead of scoring every key, making prefill, decode and training faster at long context. We propose two variants, Distilled-LSA (D-LSA), which can be applied to pre-trained models via distillation, and Native-LSA (N-LSA), which is trained natively on the next-token loss alone. On GLM-5.3-FLASH (QWEN-3.8-27B) at 10M tokens, D-LSA speeds up one attention layer by 22.7x (72.4x) in prefill and 4.6x (42.5x) in decode over its original sparse attention (full attention), while keeping similar accuracy across a variety of tasks. N-LSA matches the accuracy of full attention with a training speedup of 1.6x per layer at 128k tokens. LSA offers practical and efficient attention for models to benefit from ultra-long context.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.31585

DPPE: Rethinking Camera-Based Positional Encoding for Scaling Multi-View Transformers

Shun Kenney, Teppei Suzuki

原文摘要

The remarkable scalability of Transformers has expanded their application to 3D computer vision, where camera-aware positional encoding is crucial for providing spatial cues in multi-view geometry. Recent advancements have established the practice of using camera parameters---such as extrinsics or projection matrices---as relative positional encoding into the query, key, and value vectors of the attention mechanism. However, when scaling up the training recipe of novel view synthesis (NVS) models with the camera-based positional encoding, we observe a significant issue: model performance stagnates in the late stages of training. In this paper, we investigate the cause of the performance bottleneck when scaling up and demonstrate that storing rotation and translation given by the positional encoding in the same dimensions of the value vector causes indeterminacy in their independent identification, hindering training scalability. To address this, we propose Decoupled Pose Positional Encoding (DPPE), a novel camera-based positional encoding that explicitly decouples rotation and translation. Extensive evaluations on NVS tasks demonstrate that DPPE enables stable long-term training even in scaled-up training setup. Furthermore, it exhibits superior generalization performance in extrapolation settings, such as handling an increased number of viewpoints and zoom-in scenarios. Code is available at https://github.com/sbintuitions/dppe.

cs.CV · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2610.40982

Strong Lower Bounds for Weak Epsilon-Nets

Sujoy Bhore, Reilly Browne, Da Wei Zheng

原文摘要

Let $P$ be a finite point set in $\mathbb{R}^d$. A weak $\varepsilon$-net for $P$ with respect to convex sets is a finite set $X\subset\mathbb{R}^d$ intersecting every convex set containing at least $\varepsilon|P|$ points of $P$. Determining their optimal worst-case size was described by Alon, Kalai, Matousek, and Meshulam [Adv. Appl. Math., 2002] as "one of the truly important open problems in combinatorial geometry." We give elementary constructions which establish the first polynomially superlinear lower bounds in fixed dimension. Already in $\mathbb{R}^3$, we construct point sets for which every weak $\varepsilon$-net has size $Ω(\varepsilon^{-3/2})$. More generally, for every fixed $d\ge3$, we obtain a lower bound of $Ω_d(\varepsilon^{-γ_d})$, where $γ_d=\sqrt{d/2}+O(1)$. Consequently, no upper bound of the form $O_d(\varepsilon^{-C})$ can hold with an absolute constant $C$ which is independent of the dimension. Despite the recent progress on upper bounds, the strongest previous general lower bound remained $Ω_d(\varepsilon^{-1}\log^{d-1}(1/\varepsilon))$, due to Bukh, Matousek, and Nivasch [Israel J. Math., 2011]. To obtain our results, we apply a quadratic lift to grid-based configurations with many point-line incidences. The bounds also yield polynomial separations between fractional and integral piercing numbers of convex families, and lower bounds for Hadwiger-Debrunner numbers.

cs.CG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.03830

Identity-Consistent Multi-Pose Generation of Contactless Fingerprints

Zhiyu Pan, Zhaoliang Xu, Xiongjun Guan, Jianjiang Feng, Jie Zhou

原文摘要

Contactless fingerprint recognition has gained increasing attention due to its advantages in hygiene and acquisition flexibility. However, the absence of physical contact constraints introduces severe nonlinear geometric distortions caused by free finger poses in 3D space, resulting in a substantial cross-modal domain gap between contactless and conventional contact-based fingerprints. Existing solutions largely rely on explicit geometric correction or image enhancement, which are fragile under extreme pose variations. In this paper, we propose Identity-Consistent Multi-Pose Generation of Contactless Fingerprints (IMPOSE), a physics-inspired framework that synthesizes identity-preserving, multi-pose contactless fingerprint samples to empower recognition models. IMPOSE consists of three stages: (1) rolled fingerprint identity generation via latent diffusion with discrete codebook representations, (2) cross-modal translation from rolled to contactless modality guided by Sauvola-based local adaptive binarization as an identity anchor, and (3) physics-based multi-pose simulation through 3D finger model texture mapping and projection. The generated samples maintain strict identity consistency at the ridge topology level and spatial alignment with standard fingerprint coordinate space. Extensive experiments on the UWA and PolyU CL2CB databases demonstrate that fine-tuning fixed-length dense descriptors (FDD) with IMPOSE-synthesized data achieves state-of-the-art cross-modal matching, reducing EER to 8.74% on UWA and 2.26% on PolyU CL2CB. Synthetic data also yields consistent gains across mainstream representations including DeepPrint and AFRNet, and the hybrid strategy combining synthetic and real data achieves the best overall results. The code and generated samples are available at https://github.com/Yu-Yy/IMPOSE.

cs.CV

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04803

Extracting System Prompt of LLM Services in Complex (Agentic) Settings: Hierarchical Optimization and Reliable Evaluation Metrics

Shuaidong Li, Chenhao Lin, Xinshan Yang, Chao Shen, Yufei Chen, Zhengyu Zhao

原文摘要

System prompt is a critical intellectual property of large language model (LLM) services. To analyze its leakage, prompt extraction attacks have been extensively studied. However, we validate that existing attacks substantially degrade on recent LLM services, whose complexity has evolved in both the prompt itself and system-level constraints. Specifically, (1) system prompts become longer and more structured to contain pre-defined tool schemas, execution logic, safety guardrails, etc., and (2) only the final post-processed outputs (e.g., agent's actions) but not the raw textual response can be observed. In this paper, we address the above increased complexity by proposing Extraction with Hierarchical Optimization (ExtraHerO), which (1) decomposes the prompt extraction into a hierarchical reconstruction process with a pre-defined structural prompt template, and (2) iteratively updates the adversarial query with guidance of feedback-template differences to observe diverse outputs. In particular, we monitor the information gain of attack optimization to early stop or reset it for a good trade-off between effectiveness and efficiency. We also find that existing evaluation metrics for prompt fidelity can be misleading because they do not account for structural information. Therefore, we rely on global textual similarity and prioritize structural keywords. Moreover, we introduce a system-level metric, adversarial utility, to measure the help of prompt extraction for real-world downstream attacks. Experimental results demonstrate the consistent superiority of ExtraHerO, with average improvements of 12.5\% (fidelity), 2.3\% (functionality), and 6.0\% (adversarial utility) on chatbots, and 12.2\%, 34.3\%, and 30.5\% on agents.

cs.CR

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2609.02977

Learning Fourier-Sparse Signals and Multiband Signals

Dongrun Cai, Xue Chen, Xiaowei Shao, Yile Wang

原文摘要

A classical problem in continuous sparse Fourier transforms, which dates back to Prony's work in 1795, is to learn a $k$-Fourier-sparse signal $x(t):=\sum_{j=1}^k α_j e^{2 π\mathbf{i} f_j t}$ with arbitrary $k$ frequencies $f_1,\ldots,f_k \in \mathbb{R}$. We consider this learning problem within a fixed time window $t \in [-1,1]$ in the presence of bounded $L_2$ adversarial noise. Because the frequencies $f_1,\ldots,f_k$ can be arbitrarily close, accurately recovering individual frequencies is information-theoretically impossible even under exponentially small noise from Moitra (STOC 2015). This work therefore examines two fundamental questions: If an algorithm uses $\tilde{O}(k)$ frequencies to approximate $f_1,\ldots,f_k$ of $x$, what is the optimal covering error? And how can we efficiently learn the continuous signal $x(t)$ as a whole over the time window? Our main result is an algorithm running in time $\tilde{O}(k)$ that outputs a list $L$ of $\tilde{O}(k)$ frequencies with covering error $O(k)$ --- meaning that for most frequencies $f_j$ of $x$, $L$ guarantees $\min_{\tilde{f} \in L} |f_j-\tilde{f}| = O(k)$ --- and an reconstruction $\tilde{x}(t) \approx x(t)$ over $[-1,1]$. When constrained to outputting $\tilde{O}(k)$ frequencies, this $O(k)$ covering error is nearly optimal: Chen and Price (ICALP 2019) demonstrated two indistinguishable $k$-Fourier-sparse signals separated by distance $\tildeΩ(k^2)$ in the frequency domain, which implies an amortized lower bound of $\tildeΩ(k)$ per frequency. Furthermore, our sample and time complexities are nearly optimal, dramatically improving upon the previous best algorithm by Song, Sun, Weinstein, and Zhang (FOCS 2023), which required $\tilde{O}(k^4)$ samples and $\tilde{O}(k^{4ω})$ running time.

cs.DS

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.15044

PAT: A Personality-Driven Augmentation and Transfer Learning Framework for Depression Detection on Social Media

Shengwu Zheng, Yuzhe Ding, Xiaohui Liang, Fei Li, Chong Teng, Donghong Ji

原文摘要

Depression is a prevalent mental disorder that affects millions worldwide, with early detection crucial for effective intervention. While exist?ing methods have achieved promising results in automated depression detection, they face two key limitations: (1) noise in post screening: symptom-based semantic screening is easily misled by superficial lexical patterns, thereby introducing false-positive posts; (2) weak op?timization of the post encoder: relying solely on coarse user-level supervision signals pre?vents models from capturing depressive cues within individual posts. To address these limita?tions, we propose PAT, a Personality-driven Augmentation and Transfer learning frame?work. In the first phase, PAT optimizes a post?level encoder via a fine-grained contrastive learning objective, utilizing LLM-based data augmentation driven by classic personality the?ories to alleviate data scarcity. In the second phase, the trained post-level scorer is deployed for high-quality post filtering, while its en?coder parameters are seamlessly transferred for user-level tasks. Extensive experiments demon?strate that PAT significantly outperforms exist?ing methods. Furthermore, PAT delivers multi?granular interpretability by tracing post-level mental-state trajectories and highlighting key symptoms, thereby offering reliable diagnostic evidence for clinical practice.

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04797

Training Specialist LLMs on Your Phone: Forward-Only Optimization via Block-Spectral Coordinates

Liyan Tan, Yequan Zhao, Ziyue Liu, Yupeng Su, Ruijie Zhang, Zheng Zhang

原文摘要

Large language models (LLMs) on personal devices keep user data local, but limited memory restricts their general capability. Since most user-specific tasks do not require the breadth of frontier models, a practical solution is to adapt a shared on-device model into lightweight specialists. However, backpropagation is prohibitive in memory, and users rarely provide labeled data. We present MobileSpec, a system that trains specialists with forward passes alone and learns from automatically checked task outcomes, as in reinforcement learning with verifiable rewards (RLVR). Because forward-only updates are built from random perturbations, training is efficient only when these perturbations resemble the weight changes of RLVR, which largely preserve the singular values of pretrained weights and rotate their singular vectors. The core algorithm of MobileSpec, FuRA-S, produces such perturbations while training only the block-spectral coordinates of pretrained weights, namely the singular values of their column blocks. As we show theoretically and empirically, rescaling these blocks independently shifts their relative contributions to each weight matrix and thereby rotates its singular vectors. Since this update can be rewritten as a transformation of the layer input, FuRA-S also adapts INT4 models without modifying their weights. With specialists of only 1-5 MB that update 0.015-0.05% of the parameters, MobileSpec raises the success rate of a phone-use agent from 6-23% up to 89-100% and trains a personalized 8B model on an iPhone 16 Pro overnight. On RLVR benchmarks, FuRA-S is competitive with full-parameter evolution strategies (ES) and outperforms ES over LoRA adapters and quantized ES on INT4 models.

cs.LG · eess.SY

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04796

Construction of New Complete Complementary Codes Using Extended Boolean Functions and Permutations

Aditya Prakash, Rajen Kumar

原文摘要

Complete complementary code (CCC) is a crucial component in the design and optimization of modern wireless communication systems, particularly in addressing challenges related to interference, spectral efficiency, and signal quality. The design of CCCs is an active area of research, as it directly impacts the number of users that can be supported interference-free in the system. In the literature, CCC constructions are restricted by a set size smaller than the sequence length. In this paper, we introduce a novel construction method for CCCs based on the extended Boolean functions (EBF) that overcome existing restrictions by including a set size bigger than the sequence length. Further, the proposed construction uses functions which are new in literature and provide us with new CCCs. This is also the very first construction where row PMEPR is also bounded by $n$ for a set size in the form of power of $n$.

cs.IT

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2606.11420

Context-Aware Multimodal Claim Verification in Spoken Dialogues

Chaewan Chun, Delvin Ce Zhang, Dongwon Lee

原文摘要

Spoken factual claims often occur within multi-turn conversations, where surrounding dialogue can provide context unavailable from the claim alone. Yet most fact-checking research evaluates isolated text, leaving conversational audio under-studied. We introduce MAD2, a synthetic Multi-turn Audio Dialogues benchmark for spoken claim verification with 1,000 two-speaker dialogues, 1,230 sentence-level check-worthy candidate annotations, and approximately 10 hours of audio. We also propose calibrated multimodal fusion of a context-aware audio encoder and a dialogue-aware text model. Adding dialogue context improves verification across settings, although the gains vary by scenario. Past-only context often approaches local offline performance, suggesting its usefulness when future context is unavailable. Fusion achieves its highest mean advantage over text with full-dialogue context, but does not consistently or significantly outperform text across settings. Under full-dialogue context, exploratory subgroup results show greater performance variation across dialogue scenarios for text and fusion, but across spread styles for audio.

cs.CL · cs.SI

查看论文
arXiv ↗PDF ↗
版本更新 · v4arXiv:2603.14358

A Unified Pulse-Shaped OFDM Framework for Chirp-Domain Waveforms: Continuous-Time Modeling and Practical I/O Analysis

Yating Jiang, Hai Lin, Yi-Han Chiang, Jun Tong

原文摘要

A unified framework for chirp-domain waveforms, including orthogonal chirp division multiplexing (OCDM) and affine frequency division multiplexing (AFDM), is developed. Their continuous-time representations are shown to fall within the conventional Weyl-Heisenberg (WH) framework for multicarrier waveforms, with the root chirp as the prototype pulse. Since the root chirp has constant envelope and is transparent to subcarrier orthogonality, these waveforms can be further interpreted as pulse-shaped (PS) orthogonal frequency division multiplexing (OFDM) signals, whose power spectral density is derived analytically. The derived spectrum reveals that implementations based on the discrete affine Fourier transform rely on sub-Nyquist samples and exhibit frequency aliasing. We prove that the corresponding aliased chirps are only conditionally orthogonal, whereas sample-wise root-Nyquist pulse shaping of the discrete-time AFDM (DT-AFDM) sequence produces mutually orthogonal pulse-shaped chirps, resulting in the pulse-shaped AFDM (PS-AFDM) waveform. We then derive an exact waveform-level input-output (I/O) relation for PS-AFDM over delay-Doppler (DD) channels, showing that the effective channel at a practical receiver is generally not a superposition of pure path-wise DD components. Waveform simulations verify the derived relation to machine precision, while the conventional sequence-level I/O relation for DT-AFDM exhibits a substantial mismatch with waveform behavior for practical channels with continuous-valued delays.

cs.IT · eess.SP

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2609.23669

A Polynomial Kernel for Planar Directed Feedback Vertex Set

Zimo Sheng, Mingyu Xiao

原文摘要

The Directed Feedback Vertex Set problem (DFVS) asks whether a digraph can be made acyclic by deleting at most $k$ vertices. Whether DFVS admits a polynomial kernel parameterized by $k$ is a major open problem in kernelization, even for planar digraphs. We resolve the planar case by giving a deterministic kernel with $O(k^{66}\log^2 k)$ vertices and arcs. Our algorithm proceeds in three stages. First, we apply structural reduction rules to the input digraph, bounding the number of directed faces and some special vertices. Second, we pass to the planar dual, where vertex deletion corresponds to adding groups of reverse arcs to make each weakly connected component strongly connected. The structural bounds in the first stage yield a small retained vertex set in the dual. We then compress the dual instance by identifying vertices with the same distance records from this retained vertex set. The main technical contribution is a directed-cut argument showing that this identification preserves feasibility. Finally, we transform the polynomial-size dual instance back into an instance of Planar Directed Feedback Vertex Set via a $3$-CNF encoding and a planar graph construction.

cs.DS

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04794

Accelerating Offloading-Based MoE Inference by Interleaving Concurrent Requests

Hongseung Yu, In Gim, Kyunghan Lee, Lin Zhong

原文摘要

Expert offloading enables large mixture-of-experts (MoE) language models to run locally, but conventional batching stalls the GPU while missing experts are fetched. A popular remedy, known as prefetching, is to predict what experts will be needed soon and fetch them using idle link bandwidth. Imprecise prediction, however, wastes bandwidth and GPU memory on fetching unnecessary experts. Our key finding is that prefetching can be harmful when there is an expert cache in GPU memory. This is because as the cache includes more necessary experts, expert prediction becomes less precise. Our key insight is that proper request scheduling beats prefetching. Building on this insight, we present Bevy, an MoE inference system based on runahead inference for concurrent requests. A request whose experts are available can run ahead to later layers while other requests wait for fetches, revealing the experts it needs next. Bevy can then fetch the missing experts so identified while computing ready requests, overlapping computation and fetching without predicting expert selections. Running requests separately, however, repeats reads of shared weights in GPU memory and increases compute time. To limit this cost, Bevy's self-balancing scheduler batches ready requests at each layer and prioritizes lagging requests, adapting their grouping as expert availability changes. Across two local platforms, Bevy delivers up to 3.46$\times$ the decode throughput of llama.cpp and up to 1.55$\times$ that of the fastest prefetching baseline.

cs.DC · cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04792

Density-Aware Reward Scaling: A Reinforcement Learning Framework for Reward-Proportional Sampling

Ria Shekhawat, Abhijeet Sinha, Yingtao Zhu, Dianbo Liu

原文摘要

Reinforcement learning objectives that maximize expected return can concentrate probability mass on a small subset of high-reward outcomes, limiting diversity when multiple distinct solutions are desirable. We introduce Density-Aware Reward Scaling (DARS), a simple reward-processing mechanism that promotes reward-proportional sampling in continuous outcome spaces while retaining standard policy-gradient training. DARS estimates the local density of outcomes within each on-policy rollout group and inversely scales rewards by this density. The method requires no learned density model, auxiliary objective, critic, or modification to the underlying policy-gradient objective. In a controlled multimodal navigation task, DARS-GRPO avoids the single-mode collapse exhibited by GRPO and closely recovers the target mode allocation. Applied to text-to-image prompt adaptation, DARS-GRPO consistently improves image and prompt diversity across datasets, Policy LMs, and reward settings while maintaining good semantic alignment. These results show that density-aware reward scaling provides a lightweight way to improve coverage of diverse high-reward outcomes within existing reinforcement-learning pipelines.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v4arXiv:2609.05178

LIBERO-RECOVER: Beyond Task Success Towards Failure Recovery in Robotic Manipulation Models

Lin Liu, Zhicheng Bao, Lu Zhang, Ziying Song, Wu Yang, Yuzheng Zhuang, Shuai Tao, Wulong Liu 等 10 位作者

原文摘要

Vision-Language-Action (VLA) or World Action (WAM) models have recently demonstrated remarkable performance in robotic manipulation. On LIBERO, SOTA method have achieved nearly 100\% success rates, seemingly suggesting that the models are ready for deployment in real world. However, near perfect performance on existing benchmarks can be misleading: success under ideal conditions does not imply real world robustness. Existing benchmarks primarily evaluate task completion from predefined initial states, while real world interactions inevitably involve failures such as failed grasps, collisions, and unintended object movements. A robot must therefore not only execute tasks successfully, but also recognize and recover from failures to continue the task. Yet this capability remains largely unmeasured, revealing a critical gap between benchmark performance and real world reliability. To address this gap, we introduce LIBERO-Recover Benchmark, a large scale benchmark for failure recovery in robotic manipulation. Built upon LIBERO, we collect real execution failures from SOTA embodied models and construct 1,000+ scenarios across four recovery levels: (1) Action Retry, (2) Action Adaptation, (3) Object State Recovery, and (4) Environmental Recovery. We evaluate four core capabilities: spatial understanding, object structure reasoning, interaction understanding, and topological reasoning. As the first large-scale benchmark for embodied failure recovery, LIBERO-Recover shifts evaluation from \emph{Can the robot succeed?''} to \emph{Can the robot recover after failure?''}, promoting robust and generalizable embodied agents. The project will be avaible in \textcolor{blue}{https://liulin815.github.io/LIBERO-Recovery/}.

cs.RO

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2608.19908

A Layered Simplex Architecture for Large Alphabets

Meir Feder, Yaniv Fogel, Ruediger Urbanke

原文摘要

A learning architecture induces a prior over its expressible models. Recent work links architectural success to a broad complexity spectrum: model probabilities spanning many orders of magnitude. Understanding how depth shapes these priors and prediction may help explain the success of neural networks and transformers. We study this connection in large-alphabet prediction through the *layered simplex architecture* (LSA), which multiplies independent uniform simplex vectors coordinate-wise and normalizes. At depth zero, the prior is concentrated on the uniform distribution; at depth one, the Bayesian predictor is Laplace's rule. Each additional layer shifts prior mass toward more concentrated distributions. We derive exact Bayesian marginal likelihood and regret expressions using reusable depth-specific kernels. Mixing over depths up to a cutoff logarithmic in alphabet size yields an explicit predictor with rigorous guarantees. For targets uniform on a specified sparse support, the mixture's prior complexity is asymptotic to the support-description cost at every fixed positive KL resolution, establishing a broad complexity spectrum. We also prove a regret lower bound for every label-invariant predictor: the expected description length of the discovered symbol set minus the source's uncertainty about its identity given its size. For Zipf sources in a growing-alphabet regime, the depth mixture attains this bound to leading order. Thus, the symbol-discovery rate determines the regret exponent. Although introduced to study depth, the LSA predictor competes with established large-alphabet methods, including Good--Turing, Ristad's law and an adaptive Dirichlet mixture, while substantially outperforming classical add-constant rules on concentrated sources.

cs.IT · cs.LG · stat.ML

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2604.16767

When Misinformation Speaks and Converses: Rethinking Fact-Checking in Audio Platforms

Chaewan Chun, Delvin Ce Zhang, Dongwon Lee

原文摘要

Audio platforms have evolved beyond entertainment. They have become central to public discourse, from podcasts and radio to WhatsApp voice notes and live streams. With millions of shows and hundreds of millions of listeners, audio platforms are now a major channel for misinformation. Yet existing fact-checking pipelines are mostly designed for written claims, overlooking the unique properties of spoken media. We argue that audio misinformation is not merely textual content with transcripts: it is structurally different because it is both spoken - carrying persuasive force through prosody, pacing, and emotion - and conversational - unfolding across turns, speakers, and episodes. These dual properties introduce verification difficulties that traditional methods rarely face. This position paper synthesizes evidence across modalities and platforms, examines datasets and methods, and highlights why existing pipelines fail on audio. We argue that advancing fact-checking requires rethinking verification pipelines around the spoken and conversational realities of audio.

cs.CL · cs.CY

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2603.22910

EchoKV: Efficient KV Cache Compression via Similarity-Based Reconstruction

Shiyu Ji, Yixuan Wang, Yijun Liu, Qingfu Zhu, Wanxiang Che

原文摘要

The increasing memory demand of the Key-Value (KV) cache poses a significant bottleneck for Large Language Models (LLMs) in long-context applications. Existing low-rank KV compression methods reduce this footprint by modifying model projections, limiting the flexibility to switch back to standard full-cache inference when sufficient memory is available. In this paper, we propose EchoKV, a flexible KV cache compression framework that supports on-demand transitions from full KV caching to compressed caching. Unlike traditional compression-decompression paradigms, EchoKV utilizes a lightweight network to reconstruct the discarded KV components from a partial subset, exploiting intrinsic inter-layer and intra-layer similarities among attention heads. We further introduce a lightweight two-stage fine-tuning strategy, requiring only a few minutes on a single A100 GPU for Llama-3.1-8B. Experimental results on LongBench and RULER demonstrate that EchoKV offers a competitive quality-cache trade-off across multiple compression ratios and backbone models while preserving the throughput of full-cache inference in short-context scenarios.

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04788

Global Trace Condition for Injective Transition Graphs

Stefano Berardi, Ugo de'Liguoro, Daniel Osorio-Valencia

原文摘要

We formulate the global trace condition, GTC, for cyclic proofs as a particular case of the size-change termination condition, or SCT. We express cyclic proofs through a variant of size-change graphs, called transition graphs, whose edges are labelled by (finite) bipartite relations. We show that injective transition graphs suffice to encode cyclic pre-proofs for a variant of Brotherston and Simpson's sequent calculus with inductive definitions. The injectivity of transition graphs corresponds to restricting SCT to fan-in-free graphs. Then we prove that an essentially weaker condition implies SCT in this case, and hence GTC.

cs.LO

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2604.22507

Railway Artificial Intelligence Learning Benchmark (RAIL-BENCH): A Benchmark Suite for Perception in the Railway Domain

Annika Bätz, Pavel Klasek, Seo-Young Ham, Martin Lauer

原文摘要

Benchmark suites are crucial for advancing computer vision research, providing not only curated datasets but also standardized evaluation protocols for reproducible comparison. The railway domain lacks such resources, hindering progress in automated train operation. To address this gap, we present RAIL-BENCH, the first perception benchmark suite designed specifically for railways. RAIL-BENCH comprises five challenges - rail track detection, object detection, vegetation segmentation, multi-object tracking, and monocular visual odometry - each tailored to railway-specific characteristics and including a curated dataset from diverse real-world scenarios, evaluation metrics, and public scoreboards; we demonstrate the first three challenges in this paper. We further introduce LineAP, a novel segment-based average precision metric for rail track detection that decouples geometric accuracy from instance-level grouping and employs a track-width-based distance threshold approximating a fixed real-world distance, addressing key limitations of existing line detection metrics. Our benchmark suite is publicly available at https://www.mrt.kit.edu/railbench/.

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2609.04452

TRILOGUE: A Trilingual Spoken Dialogue Fact-Checking Benchmark with Evidence and Paired Audio

Chaewan Chun, Meruyert Aristombayeva, Jiyoung Choi, Mahjabin Nahar, Delvin Ce Zhang, Dongwon Lee

原文摘要

Modern misinformation is often heard before it is read, yet fact-checking systems are still evaluated mainly on clean written claims. Spoken dialogue remains different even when systems operate on transcripts: claims may be distributed across speakers and turns, depend on prior context, and become harder to verify when Automatic Speech Recognition (ASR) errors distort the available text. Prior spoken dialogue fact-checking resources are small, English-centric, or focused on annotation rather than end-to-end benchmarking, leaving no large multilingual benchmark with paired speech and turn-level labels. We introduce TRILOGUE (TRIlingual spoken diaLOGUE fact-checking), a large-scale trilingual benchmark of source-grounded spoken dialogues in English, Russian, and Kazakh. It contains nearly 12K dialogues, 187K turns, and 390 hours of paired audio with ASR transcripts and word-level timestamp alignments across all three languages, including nearly 5K human-recorded Russian and Kazakh dialogue files. TRILOGUE supports claim check-worthiness detection, source-article evidence retrieval, and claim verification with claim-only, gold-evidence, and retrieved-evidence inputs. Baselines show that ASR degradation and cross-lingual transfer remain challenging, especially for Kazakh, while retrieved source evidence substantially narrows the gap to gold-evidence verification.

cs.CL

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.14659

Slower Generalization, Faster Memorization: A Sweet Spot in Algorithmic Learning

Shin So, Kyelim Lee, Albert No

原文摘要

Once generalization becomes possible, does adding more data always make learning faster? In Needleman-Wunsch matrix generation, we find that small Transformers reach high exact-match validation accuracy in the fewest optimizer updates at an intermediate dataset size, while larger datasets converge more slowly. A random-suffix probe shows that, at larger dataset sizes, models fit the structured matrices before memorizing their random suffixes, suggesting that shared rule learning aids training-set fitting. To examine what makes rules easier to learn, we study elementary cellular automata (ECA), in which each binary cell updates from its own state and those of its two neighbors. These experiments identify lower-order predictive signals-output biases or predictive relationships involving only a few input bits-as a key factor in learnability. In Rule 90, finite-sample fluctuations create output biases and single-input dependencies absent from the uniform population. These signals weaken as the dataset grows. Biased training that introduces these signals accelerates generalization on an unchanged uniform validation distribution. These results support an explanation for faster convergence at intermediate dataset sizes: sampling fluctuations can facilitate rule acquisition, and larger datasets weaken this advantage.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04787

ReSI-Math: Exploring Recoverable Capability Frontiers for Efficient Reasoning Improvement

Yao Cui, Zhen Yang, Yuandong Wang, Yangfu Zhu, Tianyu Wo, Zhenzhou Shao

原文摘要

Reinforcement learning with verifiable rewards (RLVR) has become an effective approach for improving mathematical reasoning in large language models. However, a fundamental question remains: when a model fails to solve a problem, does it truly lack the underlying capability? In this work, we reveal that direct failure does not necessarily imply inability. Some reasoning failures can be recovered by exploring the model's own trajectories, exposing a recoverable capability frontier: a policy-dependent region where solutions are not reliably generated through direct decoding but can be elicited through trajectory-level search. Based on this observation, we introduce ReSI-Math, a framework that explores recoverable capability frontiers and converts recovered capabilities into data-efficient RLVR supervision. ReSI-Math uses Self-Judge-guided evolutionary search to identify recoverable failures under a fixed policy and adaptively selects frontier problems for sequential RLVR as the policy evolves. On Qwen3-VL-8B, correct-incorrect trajectory crossover produces 272 verifier-confirmed correct offspring from 621 attempts (43.8%). After three sequential one-shot RLVR stages, the MATH500 and WE-MATH average improves from 78.71 to 80.16. Under matched single-problem RLVR, random selection reaches only 53.31 versus 54.16 for frontier-guided Q1.

cs.CL · cs.AI

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04786

Near-linear-Time Sampling of the Ferromagnetic Ising Model at the Tree Uniqueness Threshold

Tianshun Miao

原文摘要

We give a near-linear-time sampler for the zero-field ferromagnetic Ising model throughout the tree-uniqueness regime, including the threshold itself. The method also gives sampling algorithms for the XY and integer Potts models. On an $n$-vertex graph with $m$ edges and maximum degree at most $Δ$, we obtain the following expected-work bounds, with total variation error at most $\varepsilon$. 1.Near-linear Ising sampling. For $Δ\ge3$, the work is $O(n+m\log^3(n/\varepsilon))$ throughout the tree-uniqueness regime, including the threshold $β=Δ/(Δ-2)$. 2.Almost-linear XY sampling. For $Δ\ge8$, we sample the XY model in $\widetilde O(m+n^{1+o(1)})$ expected work at inverse-polynomial accuracy up to and including the robust phase-transition threshold on the $Δ$-regular tree. 3.Near-linear Potts sampling. For integer $q\ge3$, we sample the $q$-state Potts model in $\widetilde O(n+q^{O(1)}m)$ expected work whenever $β-1\le 0.1(β_{\mathrm{od}}(q,Δ)-1)$, where $β_{\mathrm{od}}$ is the ordered/disordered coexistence parameter. We express the even-subgraph representation of critical Ising as a polymer model that admits near-linear-time sampling, thereby bypassing the critical slowdown of Glauber and Swendsen--Wang dynamics. Building on the low-degree approximation approach of Miao and Yin, we establish general spectral conditions for fast polymer sampling. We verify these conditions through explorations that also implement the sampler efficiently, and apply the same framework to flow representations of the XY and Potts models. We also establish polynomial-time counting and sampling for the ferromagnetic XY model at every fixed temperature.

cs.DS

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04785

Kierston & KierstonBench: Simulation of Allentown, PA and Evaluation of Language Models in Crime Investigation and Surveillance Tasks

Andrew Semenov, Svyatoslav Dorofeev

原文摘要

Large language models are candidates for collecting and analyzing records in criminal investigations and mass surveillance. Their capability and accuracy in this work cannot currently be measured. Real deployments lack the ground truth needed to score a model's conclusions, and no benchmark evaluates these capabilities in a simulated environment. We present Kierston, a city-scale simulation engine, and KierstonBench, a benchmark that tests these capabilities against known simulated events. Kierston models approximately 125,000 residents of Allentown, Pennsylvania, generating their daily activities and the telecommunications, financial, administrative and license-plate records that surveillance and investigative systems collect about them. These records derive from a common behavioral timeline and are distributed across sources representing separate providers. Models investigate through a restricted records interface, linking observations across sources without access to the underlying simulation. The benchmark comprises 32 task families covering record linkage, routine reconstruction, incident investigation and assessment of deceptive accounts. In an evaluation of sixteen language models, most could not perform records investigation and submitted identifications the records did not support. A small number of frontier models solved most of the tasks and abstained when evidence was insufficient. In some cases these models still fell for common deceptions and assumptions, and none replaced an investigator.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2609.23540

Vector Balancing in Polynomial Time

Shengtao Guo, Ethan X. Fang, Junwei Lu

原文摘要

The Komlós problem asks for signs $\varepsilon\in\{-1,1\}^n$ with $\|A\varepsilon\|_\infty$ bounded by an absolute constant whenever the columns of $A\in\mathbb R^{m\times n}$ have Euclidean norm at most $1$. We give the first polynomial-time algorithm for this problem in the unit-cost real-RAM model. Our deterministic algorithm achieves discrepancy at most $8272$. As a consequence, we obtain a deterministic $O(\sqrt t)$-discrepancy algorithm for the Beck-Fiala problem on set systems in which each element belongs to at most $t$ sets. Our algorithm rounds a fractional coloring while keeping the row sums bounded. Its key idea is to keep the number of nearly tight row constraints small relative to the number of active coordinates. We achieve this through a spectral bound based on a weighted Gram matrix of squared row entries. We maintain the bound using a cubic spectral potential and negative-curvature updates that satisfy the required linear constraints.

cs.DS · math.CO

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.04784

Optimal Approximation for Constrained Correlation Clustering via Repulsion

Yixin Cao, Ying Xu

原文摘要

The best-known approximation factor for constrained correlation clustering stood at three for nearly two decades, until a recent improvement to below $16/7$. We give a randomized polynomial-time $(2+\varepsilon)$-approximation using \emph{Repulsion}, a recursive operation that separates a metric ball from its complement. Repulsion naturally respects hard constraints and admits a short potential-based analysis; as a warm-up, radius-$2/3$ Repulsion rounds the standard metric LP into a feasible partition of cost at most three times its objective. The $(2+\varepsilon)$-approximation takes a two-stage approach, first removing must-links with arbitrarily small loss in the approximation ratio. The second stage combines lossless, constraint-aware preclustering with the bounded sub-cluster LP framework of Cao, Cohen-Addad, Lee, Li, Newman, and Vogl (STOC 2024). Repeated positive conditioning and sampling produce candidate sets, which radius-one Repulsion partitions into feasible clusters at controlled cost. We also give an approximation-preserving reduction from vertex cover to cluster deletion, a special case with only cannot-links. Apart from establishing a matching factor-two hardness threshold under the Unique Games Conjecture, it motivates our two-stage approach and Repulsion.

cs.DS

查看论文
arXiv ↗PDF ↗