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
+63已有论文修订
+93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 13:44各信源进度 →

统计口径与数据新鲜度

统计 2026-11-15 20:25 — 2026-11-16 20:25(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 13:44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6,921 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)
新论文 · v1arXiv:2611.12237

A 1.764-competitive algorithm for online fractional vertex cover on forests

Marcin Bienkowski

原文摘要

We study online fractional vertex cover on forests under edge arrival. We give a deterministic algorithm with competitive ratio at most $4 - \sqrt{5} \approx 1.764$, which simultaneously produces an online fractional matching with competitive ratio of at least $1/(4 - \sqrt{5}) \approx 0.566$. This improves on the previous best competitive ratios of $11/6 \approx 1.833$ and $5/9 \approx 0.555$, respectively.

cs.DS

新论文 · v1arXiv:2611.13039

Robo-MLT: Dual-System Robotic Manipulation for Long-Horizon Medical Laboratory Tasks

Zequan Zeng, Kai Wang, Delvin Ce Zhang, Huaping Liu, Zhipeng Yuan, Junqin Yang, YunLong Zhang

原文摘要

Reliable medical laboratory automation requires robots to complete entire workflows in protocol order while handling changes during execution. A robot may execute each motion correctly yet fail the workflow by confusing visually similar stages or stalling after human intervention. We present Robo-MLT, a dual-system framework that couples explicit workflow monitoring with learned action generation. A vision--language planner composes an ordered atomic-skill sequence from the protocol and the initial workspace observation. A multiframe execution monitor detects when the active skill has completed or become infeasible, then advances the workflow or triggers replanning. A skill-conditioned vision--language--action policy generates the manipulation trajectories, and an asynchronous runtime coordinates it with the monitor. We evaluate Robo-MLT against four policy baselines and a VLM-based monitor baseline on blood gas analysis and blood sample tube sorting, with 50 nominal configurations per method and 20 sample-removal configurations. Robo-MLT achieves task success rates of 82\% and 70\%, exceeding $π_{0.5}$ by 30 and 22 percentage points, respectively. After an operator removes the sample being handled, the system revises the remaining plan and continues processing the other samples. These results indicate that, under the evaluated conditions, explicit workflow monitoring makes long-horizon laboratory manipulation more reliable and recoverable.

cs.RO

新论文 · v1arXiv:2611.12232

ECOS: Environment-Calibrated On-Policy Self-Distillation for Language Agents

Chenhao Zhang, Zhongan Wang, Mingxuan Li, Hehe Fan, Zheng Zhu, Ming Li

原文摘要

On-policy self-distillation lets a language agent act as its own Teacher by conditioning on privileged context such as skills written from its own trajectories. Its guidance, however, is weighted by how much this context shifts token likelihoods rather than by whether it improves behavior, so errors in self-written skills return through later updates. We propose environment-calibrated on-policy self-distillation, which uses environmental outcomes to decide how much to trust the Teacher's current guidance and which self-written skills to reinforce. Command-level advantages, estimated by comparing commands executed at similar observations, scale each Teacher preference and reward reflections and skills whose induced preferences favor higher-advantage commands. Without skills at inference, 3B and 7B agents reach 90.5% and 93.3% macro success on seen ALFWorld tasks and 75.5% and 80.5% success on WebShop. On unseen ALFWorld tasks, they outperform GRPO under the same protocol by 11.0 and 7.6 percentage points, respectively. In an ablation study, supervising only executable commands rather than the Actor's full response raises 3B WebShop success from 52.3% to 75.0%.

cs.AI

版本更新 · v2arXiv:2604.11154

Environmental Footprint of GenAI Research: Insights from the Moshi Foundation Model

Marta López-Rauhut, Loic Landrieu, Mathieu Aubry, Anne-Laure Ligozat

原文摘要

New multi-modal large language models (MLLMs) are continuously being trained and deployed, following rapid development cycles. This generative AI frenzy is driving steady increases in energy consumption, greenhouse gas emissions, and a plethora of other environmental impacts linked to datacenter construction and hardware manufacturing. Mitigating the environmental consequences of GenAI remains challenging due to an overall lack of transparency by the main actors in the field. Even when the environmental impacts of specific models are mentioned, they are typically restricted to the carbon footprint of the final training run, omitting the research and development stages. In this work, we explore the impact of GenAI research through a fine-grained analysis of the compute spent to create Moshi, a 7B-parameter speech-text foundation model for real-time dialogue developed by Kyutai, a leading privately funded open science AI lab. For the first time, our study dives into the anatomy of compute-intensive MLLM research, quantifying the GPU-time invested in specific model components and training phases, as well as early experimental stages, failed training runs, debugging, and ablation studies. Additionally, we assess the environmental impacts of creating Moshi from beginning to end using a life cycle assessment methodology: we quantify energy and water consumption, greenhouse gas emissions, and mineral resource depletion associated with the production and use of datacenter hardware. Although Moshi is a specific case study, its development involves research stages common to most GenAI R&D pipelines. Our analysis therefore supports general practical recommendations for reducing compute usage and environmental impacts, while paving the way toward more sustainable MLLM research.

cs.AI

新论文 · v1arXiv:2611.12231

Towards Reliable Cloud VR over Wi-Fi 7: Leveraging MLO and TXOP Sharing

Hugo Sanz, Francesc Wilhelmi, Boris Bellalta

原文摘要

Cloud~\ac{VR} imposes stringent latency and reliability requirements due to the closed-loop interaction between \ac{DL} video streaming and \ac{UL} tracking updates. Although Wi-Fi~7 introduces \ac{MLO} to improve throughput and reduce latency, its adoption for latency-critical applications like Cloud~\ac{VR} is not straightforward. This paper investigates Cloud~\ac{VR} streaming over Wi-Fi~7 through a comprehensive cross-layer evaluation and proposes an application-aware \ac{TXS} scheduling policy that exploits the \ac{UL}/\ac{DL} coupling to jointly serve tracking and video traffic. To evaluate our mechanism, we introduce NeXus-VR, an open-source simulator that extends OMNeT++ to integrate the complete Cloud~\ac{VR} application pipeline and the Wi-Fi~7 MAC/PHY layers. Simulation results with multiple \ac{VR} users, \ac{OBSS} contention, and user mobility show that \ac{MLO} and the proposed \ac{TXS} scheduling policy provide complementary benefits. While \ac{MLO} improves transmission opportunities and robustness, the proposed Cloud \ac{VR}-aware \ac{TXS} scheduler preserves tracking freshness by reducing contention between \ac{UL} and \ac{DL} traffic. Together, they increase Cloud~\ac{VR} capacity by up to 2.25$\times$ over conventional single-link operation while consistently reducing latency and improving application-level quality of experience.

cs.NI

版本更新 · v2arXiv:2609.08277

Adaptively Incorporating Directional Hints into Zeroth-Order Optimization

Alexander Ryabchenko, Jian Qian, Wenlong Mou

原文摘要

We study zeroth-order optimization of nonconvex functions with the aid of directional hints, which are cheap but potentially inaccurate approximations of the true gradient direction, given by linear subspaces at each iteration. To leverage these hints adaptively while maintaining robustness to their quality, we introduce Control-Variate Zeroth-Order Descent (CV-ZOD), a new framework that refines the classical zeroth-order gradient estimator with a control variate that can be set based on the directional hints. We first show that the oracle algorithm that optimally sets the reference vector and step size at each iteration achieves a convergence rate that interpolates between the first-order $O(1/T)$ rate and the zeroth-order $O(d/T)$ rate, depending on the quality of the hints along the trajectory. We then develop a practical variant of CV-ZOD that achieves the same oracle guarantee up to logarithmic factors, without any prior knowledge of the hint quality. We validate the method empirically on simulation-based scientific optimization tasks, demonstrating sustained progress on nonconvex landscapes where zeroth-order descent is slower and existing guided methods stall as guidance deteriorates.

cs.LG · math.OC

新论文 · v1arXiv:2611.12226

A Game-Theoretic MPC Planner for Cooperative and Adversarial Drone Coordination

Zdeněk Bouček, Miroslav Flídr

原文摘要

We present and evaluate a game-theoretic model predictive control planner for cooperative goal reaching and adversarial pursuit-evasion with drones. The planner combines 3D prediction, individual objectives, optional shared separation constraints and trajectory-tracking interfaces. A comparison of cooperative game solvers assesses their ability to find feasible trajectories that resolve conflicts. A computational study quantifies how number of drones, prediction horizon and move blocking affect computation time and numerical acceptance. Move blocking holds control inputs constant over groups of prediction steps, reducing the number of independent control variables and lowering computation time. Timing measurements identify the practical limits of real-time planning in the tested configurations. PyBullet simulations demonstrate cooperative goal reaching and pursuit-evasion. A laboratory flight complements the simulations by validating the trajectory-tracking interfaces on physical drones. The planner can serve as a basis for further evaluation and controller development.

cs.RO · eess.SY

新论文 · v1arXiv:2611.12225

A Cost That Practice Accuracy Hides: Worked Solutions in Self-Generated Training Data and the Compositional Use of New Rules

Jiantong Xu

原文摘要

A worked solution can make self-generated training targets shorter and cheaper, but accuracy on the practiced operation may hide how well a model can reuse newly learned rules. We study this choice in a 2B-parameter reasoning model learning two invented quantitative rules. The model generates targets from a rule sheet alone or from the sheet plus a program-written worked solution; targets are accepted for final-answer correctness, and the support is removed for fine-tuning and evaluation. Training questions request forward use in one segment; composition requires sequential reuse of the same rule across two segments. After six passes, using a program-checked numerical signature, sheet-only targets lead by 24.3 percentage points (95% item-bootstrap CI 14.6-34.0) on composition in a prospective test. On unseen argument values, observed forward accuracy is identical (87.5% each), but composition accuracy is 58.3% versus 23.6%, a 34.7-point gap (25.7-43.8). An independently sampled second target-bank pair repeats the direction (+14.6 points, 2.1-26.0). Ten passes on the shorter worked-solution targets, approximately matching the domain tokens of six sheet-only passes while changing other training quantities, narrow the gap to 16.3 points (6.3-26.7) in the first bank pair and 6.2 points (-6.3 to 18.8) in the second, where the pre-specified test does not hold. At six passes each, the worked-solution policy uses fewer capture tokens and less update time, yet the trained models use 2.6 times as many generated tokens per composition success under the numerical signature (2.1 times in the second bank pair). Choosing a capture context therefore requires testing how the new rules are reused, alongside the inference cost of those uses, rather than relying on forward accuracy alone.

cs.SE · cs.CL · cs.LG

版本更新 · v2arXiv:2611.08864

Adaptive Scheduling Pipeline for Multi-Instance Asynchronous Reinforcement Learning

Salah Chikhi, Luis Zuin, Entong Li, Zeng Li

原文摘要

We present the Adaptive Scheduling Pipeline (ASP), an offline rollout scheduler for multi-instance asynchronous reinforcement learning (RL) of large language models (LLMs). ASP jointly reduces three sources of rollout underutilization - padding bubbles from intra-minibatch length heterogeneity, skewness bubbles from long-tail outputs, and sorting bubbles from fluctuations in training-queue occupancy induced by output-length-sorted dispatch - through a single offline preprocessing pass. This enables efficient multi-instance rollout that overcomes the early scaling plateau of single-instance inference while reducing fluctuations in the supply of generated samples to the trainer. We integrate the scheduler into an asynchronous architecture that decouples rollout from training while keeping policy lag observable and bounded. Within our asynchronous architecture, ASP's scheduling reduces mean iteration time by 10.9%. The complete system improves end-to-end throughput by up to 3.22 times and reduces iteration time by up to 76.8% relative to MindSpeed-RL, while reducing iteration time by up to 62.5% relative to verl. It also scales more robustly with additional generation resources than these baselines and shows no visible convergence degradation in the measured bounded-lag regime. ASP has since been used in an internal telecom-domain RL post-training workflow, reducing the reported training period from more than three months to two months and producing a domain-enhanced model for knowledge answering and tool calling.

cs.DC · cs.AI · cs.LG

新论文 · v1arXiv:2611.12223

Mitigating the BICM Loss in MIMO Receivers with Correlation-Informed Logits

Christoph Studer

原文摘要

Bit-interleaved coded modulation (BICM) is widely used in multiple-input multiple-output (MIMO) wireless systems. However, separating MIMO data detection from channel decoding and only exchanging per-bit soft information discards correlation information among the bits within each transmit vector. We show that the resulting information loss limits the performance of practical one-shot receivers. Even though iterative detection and decoding (IDD) can recover most of this loss, IDD significantly increases receiver complexity and latency. In order to reduce this loss, we propose COrrelation-Informed Logits (COIL), a soft-output format that combines per-bit soft information with pairwise coupling logits for correlated bits. We show how to incorporate these coupling logits into a conventional belief-propagation-based channel decoder, which enables correlation-aware decoding in a one-shot receiver without IDD. We evaluate COIL in a MIMO-BICM system with a conventional low-density parity-check code. For a 4x4 MIMO orthogonal frequency-division multiplexing (OFDM) system, a one-shot linear minimum mean-square error (LMMSE) receiver with COIL outperforms a conventional one-shot unbiased LMMSE receiver by 1.3dB and operates within 0.8dB of four-iteration IDD, while requiring only a single pass through the data detector.

cs.IT · eess.SP

新论文 · v1arXiv:2611.12222

PARIS: A multi-agent LLM framework for end-to-end, PRISMA-guided systematic literature reviews with pre-synthesis corpus quality assessment

Christophe Cruz, Hussam Ghanem, Samir Jabbar, Ait Kbir, Hocine Cherifi

原文摘要

Systematic literature reviews (SRs) remain slow and labor-intensive, and most automation tools address single stages of the workflow. We present PARIS (PRISMA Agentic Research Intelligence System), an open-source framework in which a conductor agent running in Claude Code orchestrates 15 specialized large language model (LLM) agents, from scoping a research question to exporting a manuscript draft. Agents are specified declaratively, coordinate exclusively through persisted artifacts that the conductor validates before every transition, and delegate retrieval, bibliometric analysis and scoring to deterministic Python utilities. Three human checkpoints gate study inclusion, corpus adequacy and the synthesis outline. We also introduce QualityScore, a pre-synthesis index that aggregates normalized network, bibliometric, semantic and process sub-scores into a 0-100 grade. Two case studies illustrate both: AI chatbots and university students' learning (388 records, 100 included) and AI-assisted SR methods (131 records, 123 included, QualityScore 69.1 recomputed with a proxy network sub-score). They show end-to-end feasibility across domains and the diagnostic use of QualityScore: in the first run, corpus improvements chosen at the corpus-adequacy checkpoint raised the score from 74.6 to 81.9 and the number of pre-2023 studies from 1 to 19. Audits of both runs exposed failure modes: incomplete retrieval, counts that drifted between agents, unrecorded or delegated checkpoint decisions and a grade that depends on how citation data are handled. PARIS fully supports 8 of the 27 PRISMA 2020 items and partially supports 11. Output quality has so far been judged only by an internal reviewer agent; independent validation against human reviews is the next step.

cs.AI

版本更新 · v2arXiv:2610.35374

GIFT: Representation Geometry Matters for Single-Domain Generalized Object Detection

Yin Zhang, Yaoyue Zheng, Yongqiang Zhang, Bogdan Raducanu, Dan Liu

原文摘要

Single-Domain Generalized Object Detection (S-DGOD) aims to generalize from a single source domain to unseen domains under distribution shifts. Existing efforts mainly focus on simulating unseen domains or suppressing domain-specific factors. However, they largely overlook the intrinsic generalization capability embedded in Vision Foundation Models (VFMs). In this work, we show that unconstrained fine-tuning disrupts the geometric structure of pre-trained representations, which encodes domain-invariant semantics. Motivated by this, we propose Geometric-Invariance Fine-Tuning (GIFT), a geometric-aware fine-tuning method for VFMs. Specifically, we introduce a series of learnable Householder reflections to the pre-trained weights to encourage geometry-consistent adaptation under domain shift. In addition, we propose a lightweight channel-wise scaling to facilitate effective task-specific adaptation. Extensive experiments demonstrate that GIFT consistently outperforms state-of-the-art (SOTA) approaches on two S-DGOD benchmarks. Notably, by introducing fewer than 1% additional trainable parameters into the frozen backbone, the proposed method achieves improvements of +7.9% and +2.9% in mPC over the previous SOTA on the Cityscapes-C and Diverse Weather Dataset, respectively. This finding highlights that preserving representation geometry is crucial for robust cross-domain generalization, establishing GIFT as a principled and efficient solution to domain shift.

cs.CV

新论文 · v1arXiv:2611.12219

An Empirical Evaluation of Quantum Kernel Stability in Network Intrusion Detection

Sreejonty Chakraborty, Adrita Rahman Tory, Muhammad Aminur Rahaman, Khondokar Fida Hasan

原文摘要

Quantum kernel methods have been proposed for network intrusion detection, yet their performance limitations are often attributed to noise in current quantum hardware. This study evaluates that assumption through a controlled analysis of Pegasos-QSVC on the NSL-KDD and ToN-IoT datasets. We perform 56 quantum experiments using one exact noiseless statevector backend and three noise-simulated superconducting backends while keeping preprocessing, PCA features, data splits, and classical baselines identical. XGBoost achieves accuracies of 98.94% on NSL-KDD and 99.41% on ToN-IoT, compared with 91.80% and 82.98% for Pegasos-QSVC. The quantum model also exhibits substantially greater seed-dependent variability. Eight runs collapse into degenerate classifiers, including four recall-collapse cases and four trivial-positive classifiers. Notably, two failures occur on the noiseless backend, demonstrating that hardware noise is not a necessary condition for classi-fier collapse. Kernel Gram matrix analysis further reveals low effective rank and severe ill-conditioning even in the absence of noise, indicating intrinsic fragility in the learned quantum representation. These results show that average accuracy alone can conceal operationally unacceptable behavior and highlight the importance of reporting worst-case performance, seed-level variability, class-specific metrics, and explicit degeneracy tests in quantum intrusion detection studies.

cs.LG

新论文 · v1arXiv:2611.12210

Tight Competitive Bounds for Online Matching on Forests

Bartłomiej Bosek, Grzegorz Gutowski, Paweł Putra, Anna Zych-Pawlewicz

原文摘要

We study the online matching problem, its fractional relaxation, and the fractional online vertex cover problem on forests whose edges arrive one at a time in an arbitrary order. Matching decisions are irrevocable, and vertex-cover weights can only increase over time. For online matching, we present a randomized algorithm with expected competitive ratio $4/7$ against an oblivious adversary, improving the previous $5/9$ guarantee. We also present a deterministic algorithm that simultaneously maintains a $4/7$-competitive fractional matching and a $7/4$-competitive fractional vertex cover. This deterministic algorithm improves the previous $5/9$ guarantee for fractional matching and the $11/6$ guarantee for fractional vertex cover. We prove that $4/7$ is optimal for both randomized and fractional matching, even on growing trees of maximum degree four.

cs.DS

新论文 · v1arXiv:2611.12206

A Geometric Invariant of Attention Predicts Head Specialization from Weights Alone

Miguel P. Bento

原文摘要

Transformers are built with continuous symmetries in the attention mechanism. We derive $ρ$, a scalar invariant that measures each head's capacity to couple to rotary positional encodings. This framework allows us to analyze heads without forward passes, opening a new research direction into data-independent analysis of modern LLMs. Empirically, we find that $ρ$ predicts induction heads, the most-studied circuit type in mechanistic interpretability, with Spearman correlation between $-0.55$ and $-0.74$ across four bias-free RoPE models spanning $135$M to $7$B parameters. As a preliminary proof of concept, we use $ρ$ to apply YaRN to only the top-50% positional heads, which degrades perplexity less than uniform application. Code is available at https://github.com/MPedraBento/rho-invariant.

cs.LG · cs.CL

新论文 · v1arXiv:2611.12205

Graph-Based RL for Sequential Rendezvous Planning with Multiple Patrolling USVs

Seung Woo Han, Hyeon Bin Yeo, Seung Heon Oh

原文摘要

In patrol missions involving multiple unmanned surface vehicles (USVs), a designated command USV can support fleet management by sequentially rendezvousing with each patrolling USV. These operations require motion-feasible route planning and priority-aware rendezvous sequencing, with the objective of minimizing mission completion time and total tardiness relative to priority-dependent due dates. We propose a framework combining grid-based path planning, surrogate travel-time prediction, and graph-based reinforcement learning. The path planner accounts for ocean currents and approximates turn-rate constraints through multi-stage maneuvers, with path feasibility empirically validated through trajectory-tracking simulations. For rendezvous sequencing, a Feasible Transition Graph (FTG) represents candidate rendezvous events and temporally feasible transitions, enabling a graph neural network to exploit spatiotemporal rendezvous relationships for policy learning. A surrogate predicts travel times from compact current and motion features to avoid repeated path searches during FTG construction and policy training. Evaluation using grid-based path planning shows statistically significant average objective reductions of 10.5%, 3.5%, and 1.5% compared with greedy, MLP-RL, and 2-step lookahead greedy, respectively. The policy achieves these improvements while requiring approximately 1/35 of the latter's computation time. It maintains its advantage over MLP-RL under unseen priority-tier settings, with larger gains as fleet size increases.

cs.RO · cs.AI

新论文 · v1arXiv:2611.12202

Co-Creating Neurorehabilitation Technology Alongside Lived Experience Experts: A Participatory Design Workshop

Irene L. Y. Beck, Salvatore L. Cucinella, Johannes B. J. Bussmann, Natacha Léon, José López-Sánchez, Francesca Lunardini, Katherine L. Poggensee, Cristina Vázquez-González 等 10 位作者

原文摘要

Involving patients' lived experience (LE) early in neurorehabilitation technology design can improve the relevance and effectiveness of solutions, yet LE experts are often involved only when finished prototypes are tested. We present and evaluate a participatory design (PD) workshop for early LE involvement, examine how LE contributions influence design processes and outcomes, and provide recommendations for similar approaches. We conducted a PD workshop at the 6th International Conference on NeuroRehabilitation. Five people who have had a stroke (LE experts) and 18 workshop attendees (including the LE experts' caregivers) were organized into five groups. Following the Discover, Define, and Develop phases of the Double Diamond model, each group used Journey Maps and Data Collection Notebooks to co-create a solution for an objective specified by an LE expert. Journey Maps were thematically analyzed, and group audio transcripts underwent structured extraction using a large language model. LE input shifted discussions from discipline-based assumptions toward priorities grounded in lived experience, including fear, confidence, sensory discomfort, fatigue, aesthetics, and independence. Narrow objectives, such as opening a blister pack, produced task-specific solutions, whereas broader objectives generated device concepts, including sensory feedback for cycling and a stimulation-equipped cooking splint. Satisfaction was high among attendees and LE experts. All workshop tools and protocols are shared to support adoption. The workshop offers a practical foundation for involving LE experts early in neurorehabilitation technology development. Future iterations should better capture psychological and social dimensions, allow more reflection, and include the full range of relevant stakeholders through sustained transdisciplinary engagement between researchers and the LE community.

cs.HC · cs.RO

新论文 · v1arXiv:2611.12200

Learning to Act with World Models by Valuing What Comes Next

Chenghao Gao, Zhi Zhang, Dian Jin, Yihe Pan, Doudou Zhou

原文摘要

A sparse terminal reward tells an agent little about its steps. Much of the missing information is implicit in the environment's dynamics, because each state transition can change the conditions for success long before the reward arrives. A world model trained on real transitions can learn to predict what each action changes, but a predicted observation does not by itself show whether success has become more likely. With a binary terminal reward, PPO's critic makes this judgment by estimating the current policy's probability of success, but only for states the agent reaches. Our method, WM-PPO, elegantly brings the two together. For each action the policy samples, executed or not, the world model predicts the next observation and the critic values the predicted successor. In the policy channel, comparisons among these values give the policy auxiliary advantages; in the value channel, their mean gives the critic an auxiliary target. A shared gate weights both channels according to how reliably comparisons with similar margins predicted real outcomes earlier in training. The policy, critic and world model all learn from the agent's own interaction, with no teacher or privileged information, and only the policy is used at inference. With a comparable number of real environment steps, WM-PPO raises success from PPO's 63.5% to 87.5% on ALFWorld and from 39.3% to 52.9% on ScienceWorld. It matches the strongest PPO refinements we tested on ALFWorld and exceeds the best of them by 6.5 points on ScienceWorld. The trained policy more often completes prerequisite steps before placing an object or committing to an answer.

cs.AI

版本更新 · v2arXiv:2606.02248

Geometric Latent Reasoning Induces Shorter Generations in LLMs

Shashi Kumar, Yacouba Kaloga, Petr Motlicek, Ina Kodrasi, Andrea Cavallaro

原文摘要

Large language models solve complex problems by generating lengthy chains of explicit reasoning tokens. While effective, this makes reasoning expensive, length-sensitive, and constrained to (discrete) natural language. While latent reasoning offers a continuous alternative, determining useful structures for intermediate latent states is an open challenge. In this paper, we formulate latent reasoning as a geometric path-approximation problem within the model's pretrained token-embedding space. We introduce Geometric Latent Reasoning (GLR), which uses a lightweight transition head to predict iterative direction updates in embedding space. Using textual chain-of-thought traces as anchors, GLR learns to approximate discrete reasoning trajectories while permitting continuous deviations from exact token embeddings. Evaluations on mathematical reasoning benchmarks using Qwen3 models reveal an emergent phenomenon: geometric latent reasoning induces substantially shorter generations without an explicit length objective. By replacing early explicit reasoning with continuous latent steps, models often reach correct answers using substantially fewer total generation steps. These findings suggest that continuous trajectories act as compact intermediate reasoning states, exposing a new tradeoff between latent computation budget, output length, and accuracy. Our code is publicly available at: https://github.com/idiap/GLR. The project page is available at: https://glr-llm.github.io/.

cs.CL

版本更新 · v2arXiv:2610.21743

Pairwise AUC Optimization Needs Corrective Power: A Unified View

Jia Chen, Zhiyong Yang, Shilong Bao, Qianqian Xu, Qingming Huang

原文摘要

Direct pairwise AUC-surrogate training is attractive because it trains toward the ranking metric used at test time, especially under class imbalance. This creates a direct alignment between the training objective and the evaluation metric, but such metric alignment alone does not guarantee that stochastic training remains corrective. We study this gap through Corrective Power: whether pairwise mini-batch updates produce stronger and more reliable improvements for more severe positive-negative ranking errors. Our main observation is that direct pairwise training can enter regions where severe errors remain common but no longer produce useful corrective motion. In such regions, even the mini-batch noise that would normally help SGD move along the steepest severe-error correction direction can become much weaker. In this paper, we seek a theoretical explanation for this by exploring the degenerate property of the U-statistics formed from the pairwise gradients therein. Our results provide a mechanistic explanation for several empirical patterns: (a) squared-hinge loss is often an effective AUC surrogate; (b) cross-entropy warm-up can help models that enter the AUC phase from a poor starting region; and (c) parameter-efficient fine-tuning from a pretrained model can reduce that need by starting closer to a favorable ranking geometry. Experiments on 8 CV tasks and 8 NLP tasks, using ResNet50, DenseNet121, DistilBERT-base, and Qwen2.5-1.5B-Instruct, support these predictions.

cs.LG

新论文 · v1arXiv:2611.12198

Instance-Optimality of Single-Pair Personalized PageRank

Christian Bertram, Mads Vestergaard Jensen, Mikkel Thorup, Hanzhi Wang, Shuyi Yan

原文摘要

We consider discounted random walks in a directed simple graph $G=(V,E)$. Such a walk terminates at each step with a constant probability $α\in (0,1)$ and otherwise continues to a uniform random out-neighbor. Given a pair $(s,t)\in V^2$, we denote by $π(s,t)$ the probability of such a discounted random walk starting at $s$ terminates at $t$. This probability is referred to as the Personalized PageRank (PPR) of $t$ with respect to $s$. The problem of estimating $π(s,t)$ within a constant multiplicative factor with constant success probability for a given vertex pair $(s,t)$ is called the single-pair PPR estimation problem. We investigate the instance-optimality of single-pair PPR estimation in directed graphs. We present a new algorithm that aims at instance-optimal running time up to polylogarithmic factors. In other words, no correct algorithm can run faster than our algorithm on any valid graph by more than a polylogarithmic factor. We show that our algorithm is instance-optimal in the standard adjacency-list query model if $n$ is not known (as is often the case for most of the large networks to which PPR is applied). When $n$ is known, our algorithm is still instance-optimal for sparse graphs. For denser graphs, when $n$ is known, we show that no algorithm can achieve general instance-optimality. For the multiplicative approximation of PPR, we also improve the worst-case complexity from $O(mn\log n)$ to the near-optimal $O(m\log n)$. This is not only for the single-pair discussed above, but also for the single-source and single-target variants of the PPR estimation problem.

cs.DS

新论文 · v1arXiv:2611.12197

TrajLift: Lifting Memory Recall Trajectories with Latency-Modulated Heat Diffusion

Jiawen Kang, Jinchao Li, Mingyu Cui, Junan Li, Xixin Wu, Helen Meng

原文摘要

Verbal memory retrieval carries rich diagnostic information for Alzheimer's disease through the order, hesitation, and categorical organization of recalled words, yet computational modeling of this retrieval process remains largely unexplored. A key challenge is that timing and hierarchical structure are diagnostically entangled: the diagnostic meaning of retrieval timing depends on its hierarchical context. We propose TrajLift, a spectral framework that models verbal recall as timed trajectories on a hierarchical graph. Our central mechanism is latency-modulated heat diffusion on the semantic hierarchy, a unified operator that lifts each recalled word into a representation of both multi-scale hierarchical structure and continuous retrieval timing. Formal analysis shows that the resulting representation is first-order separable into structural and temporal components with spectral selectivity across hierarchical levels. On synthetic benchmarks requiring structure-time interaction, TrajLift stays near-perfect on deep hierarchies where all baselines fail. On real-world clinical data, we show that TrajLift outperforms diverse baselines by jointly modeling timing and semantic structure, with gains that grow as retrieval timing becomes more informative.

cs.CL · cs.AI · q-bio.NC

版本更新 · v2arXiv:2602.13930

MamaDino: A Hybrid Vision Model for Breast Cancer 3-Year Risk Prediction

Ruggiero Santeramo, Igor Zubarev, Florian Jug

原文摘要

Breast cancer screening programmes increasingly seek to move from one-size-fits-all intervals to risk-adapted and personalized strategies. Deep learning (DL) has enabled image-based risk models with stronger 1- to 5-year prediction than clinical models, but leading systems (e.g., Mirai) typically use convolutional backbones, high-resolution inputs (>1M pixels) and simple multi-view fusion, with limited explicit modelling of contralateral asymmetry. We hypothesised that combining complementary inductive biases (convolutional and transformer-based) with explicit contralateral asymmetry modelling would allow us to match state-of-the-art 3-year risk prediction performance on lower-resolution mammograms, suggesting that using less detail in a more structured way can recover state-of-the-art accuracy. We present MamaDino, a mammography-aware multi-view attentional DINO model. MamaDino fuses frozen self-supervised DINOv3 ViT-S features with a trainable CNN encoder at 512x512 resolution and aggregates bilateral breast information via a BilateralMixer to output a 3-year breast cancer risk score. We train on 53,883 women from OPTIMAM (UK) and evaluate on matched 3-year case-control cohorts: an in-distribution test set from four screening sites and an external out-of-distribution cohort from an unseen site. At breast level, MamaDino achieves discrimination comparable to Mirai on internal and external tests while using ~13x fewer input pixels. Adding the BilateralMixer improves patient-level discrimination to AUC 0.736 (vs Mirai 0.713) in-distribution and 0.677 (vs Mirai 0.676) out-of-distribution, with overlapping confidence intervals. Performance was evaluated across age, ethnicity, scanner manufacturer, tumour type and grade. These findings suggest that complementary inductive biases and explicit contralateral modelling can achieve Mirai-comparable discrimination at lower resolution.

cs.CV · cs.LG

新论文 · v1arXiv:2611.13036

CapBuddy: Efficient Physical Memory Management for Capability-based Resource-controlling Microkernel

Zhuang Lu, Guangtao Zhu, Shijun Zhao, Qianying Zhang, Lixin Chen, Zekun Xu, Wu Luo, Rui Hou

原文摘要

Capability-based resource-controlling microkernels securely multiplex hardware by exposing low-level resource abstractions guarded by capability systems. Such a design helps provide an application-specific resource management policy to achieve high performance without sacrificing security. However, the misuse of capability-protected secure abstractions can introduce performance overhead in user-level resource management, affecting the runtime performance of dynamic applications in emerging scenarios. In this paper, we focus on the secure abstraction of physical memory in microkernels with capability-based resource management delegation to investigate an appropriate design for dynamic memory management. We propose CapBuddy, an approach to building an efficient buddy system on capability-protected secure abstractions of memory, to bypass the performance overhead of the capability system by offloading functionalities that can run offline from the critical path of serving a memory request. We implemented a prototype system on the seL4 library OS, and the evaluation results show that our approach improves the performance of the seL4 user-level buddy system by 20 to 54.5 times and achieves the same order of magnitude as Linux. Further burst experiments evaluate CapBuddy's scalability and show that it maintains low median allocation latency as the number of concurrent clients increases.

cs.OS

新论文 · v1arXiv:2611.12196

The Deception Archive as a repository of standardized datasets for large-scale verbal deception detection research

Caterina Borgese, Riccardo Loconte, Lucca Pfründer, Bennett Kleinberg

原文摘要

Computational advances in natural language processing and machine learning have substantially expanded the use of automated approaches in verbal deception research. However, existing verbal deception datasets remain fragmented across disciplines, as they are often difficult to identify, access, and reuse, limiting cross-study comparisons and the development of robust computational models. We present the Deception Archive, a curated and standardized repository of publicly available verbal deception datasets built upon a large-scale systematic review. Eligible datasets were retrieved, manually inspected, and standardized following a structured data curation workflow. The current repository comprises 41 publicly available datasets containing 387,995 truthful or deceptive statements. The archive represents a heterogeneous repository (e.g., in data size, topic of investigation, ground-truth, etc.) and a living resource for integrating newly available deception datasets over time. By providing centralized access to standardized datasets and accompanying metadata, the Deception Archive facilitates dataset discovery, cross-dataset validation, and reproducible evaluation of computational models.

cs.CL

版本更新 · v2arXiv:2412.17330

EcoSearch: A Constant-Delay Best-First Search Algorithm for Program Synthesis

Théo Matricon, Nathanaël Fijalkow, Guillaume Lagarde

原文摘要

Many approaches to program synthesis perform a combinatorial search within a large space of programs to find one that satisfies a given specification. To tame the search space blowup, previous works introduced probabilistic and neural approaches to guide this combinatorial search by inducing heuristic cost functions. Best-first search algorithms ensure to search in the exact order induced by the cost function, significantly reducing the portion of the program space to be explored. A key limitation of existing best-first search algorithms is that they slow down over time: the delay between outputting successive programs grows with the number of programs already generated. We ask whether best-first search algorithms with constant delay exist. We design Eco Search, a new best-first bottom-up search algorithm. It combines the cost tuple representation of Bee Search with the per non-terminal data structures of Heap Search, thanks to a new frugal expansion strategy, and it uses bucket queues instead of priority queues, which is possible because the costs stored in each queue always lie in a window of constant width. We prove that Eco Search is the first constant-delay best-first search algorithm for pre-generation cost functions. Experimentally, we evaluate Eco Search on two classic program-synthesis benchmarks (DeepCoder and FlashFill) and observe that Eco Search solves 12% to 70% more tasks than its predecessors, and reaches the number of tasks they solve 1.6 to 3.2 times faster. This shows that constant delay can be achieved for best-first search in program synthesis, and that it makes a difference in practice.

cs.LG · cs.AI · cs.PL

版本更新 · v3arXiv:2605.21706

Latent-space Attacks for Refusal Evasion in Language Models

Giorgio Piras, Raffaele Mura, Fabio Brau, Maura Pintor, Luca Oneto, Fabio Roli, Battista Biggio

原文摘要

Safety-aligned language models are trained to refuse harmful requests, yet refusal behavior can be suppressed by steering their internal representations. Existing methods do so by ablating a refusal direction from model activations, aiming to remove refusal from the model's residual stream. Despite their empirical success, these methods lack a principled account of the latent-space transformation they induce and why it suppresses refusal. In this work, we recast refusal suppression as a latent-space evasion attack against linear probes trained to separate refused from answered prompts. Under this view, prior work's difference-in-means direction naturally defines such a probe, and its ablation is exactly a projection onto its decision boundary, i.e., a minimum-confidence evasion attack. This perspective not only explains the empirical success of prior work but also admits a key limitation: evasion stops at the decision boundary, motivating the need to push representations further into the compliant region, i.e., where the model answers. We leverage this by proposing a Controlled Latent-space Evasion attack that projects representations past the boundary with an optimized confidence. Across 15 instruction-tuned, multimodal, and reasoning language models, we achieve higher attack success rates than existing refusal-ablation baselines, also outperforming jailbreak attacks.

cs.AI

新论文 · v1arXiv:2611.12189

Computational Structure of Pairwise Housing Connectedness Networks

Seo In Jang, Seung Ho Choi, Ji-Eun Shin, Hyunwoo Lee

原文摘要

Directed housing-submarket networks combine common market movement with heterogeneous bilateral relations, and different node-level measures need not yield distinct empirical rankings. We construct a positive pairwise net-connectedness network from the generalized forecast error variance decomposition (GFEVD) of a vector autoregression (VAR) and use it to compare complementary node positions. Net directional connectedness (NET), PageRank, and Viral Centrality summarize net directional balance, recursively weighted incoming position, and normalized multistep reach. When bilateral net differences are nonzero, the network is a weighted tournament. A least-squares edge-flow decomposition separates the pairwise-net matrix into a NET-equivalent gradient component and a NET-orthogonal residual, while a monotone-allocation condition establishes exact PageRank-indegree correspondence in a transitive tournament. In Seoul's 25-district network, the gradient component accounts for 70.9% of the squared Frobenius norm, leaving 29.1% in the residual component. PageRank is strongly aligned with indegree (Spearman $ρ=0.987$), and Viral Centrality with outdegree ($ρ=0.935$); the PageRank alignment exceeds the range of independently weighted simulations with comparable tournament consistency and edge-weight variation. Three common-component treatments reduce total connectedness from 81.98% to 50.60-52.00%, a 36.6-38.3% reduction relative to the raw system, and reverse 97-98 of 300 pairwise net directions. Thus, both the magnitude and the apparent direction of connectedness depend materially on the treatment of citywide common movement. Songpa's positive NET position remains prominent across common-component treatments and dynamic specifications and ranks first in both chronological subsamples.

cs.CE · cs.SI

版本更新 · v2arXiv:2603.23439

SIGMA: A Physics-Based Benchmark for Gas Chimney Understanding in Seismic Images

Bao Truong, Quang Nguyen, Baoru Huang, Jinpei Han, Van Nguyen, Ngan Le, Minh-Tan Pham, Doan Huy Hien 等 9 位作者

原文摘要

Seismic images reconstruct subsurface reflectivity from field recordings, guiding exploration and reservoir monitoring. Gas chimneys are vertical anomalies caused by subsurface fluid migration. Understanding these phenomena is crucial for assessing hydrocarbon potential and avoiding drilling hazards. However, accurate detection is challenging due to strong seismic attenuation and scattering. Traditional physics-based methods are computationally expensive and sensitive to model errors, while deep learning offers efficient alternatives, yet lacks labeled datasets. In this work, we introduce \textbf{SIGMA}, a new physics-based dataset for gas chimney understanding in seismic images, featuring (i) pixel-level gas-chimney mask for detection and (ii) paired degraded and ground-truth image for enhancement. We employed physics-based methods that cover a wide range of geological settings and data acquisition conditions. Comprehensive experiments demonstrate that SIGMA serves as a challenging benchmark for gas chimney interpretation and benefits general seismic understanding.

cs.CV

新论文 · v1arXiv:2611.12187

Harnessing Vision-Language Models for Federated Unsupervised Visual Recognition

Jian Liang, Kuangpu Guo, Ran He, Tieniu Tan

原文摘要

Federated learning (FL) is a promising research paradigm that enables collaborative model training across distributed clients without data sharing. To handle unlabeled data in local clients, existing unsupervised FL research has primarily focused on tasks like clustering and self-supervised representation learning, leaving the critical problem of unsupervised visual recognition largely unexplored. To bridge this gap, we introduce FedUR, the first framework designed for unsupervised recognition by leveraging the zero-shot capabilities of pre-trained vision-language models like CLIP. Specifically, FedUR addresses data heterogeneity through a pseudo-labeling mechanism with a class-aware data selection strategy. This strategy first estimates a global pseudo-label distribution, then allocates pseudo-labeled data to clients accordingly. To handle the uncertainty of conventional positive pseudo-labels, FedUR further incorporates class weights and a negative pseudo-labeling strategy. Furthermore, FedUR proposes a personalized prompt aggregation protocol where general visual prompts are aggregated on the server, and personalized text prompts encoding local knowledge are retained on clients. Extensive experiments demonstrate that FedUR consistently outperforms baseline methods under both label-level and feature-level distribution shifts. We believe this work paves the way for further advances in federated unsupervised recognition. Code will be available at {github.com/tim-learn/FedUR/}}.

cs.LG · cs.CV