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

+1072新增论文
+845其中 AI 相关
+94已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-10 06:03 · 全站最近入库：2026-11-09 13:22各信源进度 →

统计口径与数据新鲜度

统计 2026-11-09 06:20 — 2026-11-10 06:20（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：2026-11-10 06:03最近发现新增版本：2026-11-09 13:22库内论文最新更新时间：2026-11-07 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。

18,172 已收录论文15,506 AI 相关4,527 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 13,660 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-10 06:02（北京时间）
版本更新 · v2arXiv:2608.06004

Do Tabular Foundation Models Agree with Themselves?

Christian Klötergens, Vijaya Krishna Yalavarthi, Lars Schmidt-Thieme, Tom Hanika

原文摘要

Tabular Foundation Models (TFMs) are the state of the art for many tabular prediction tasks. They are transformers that approximate the Bayesian posterior predictive distribution under a pretraining prior. TFMs become multivariate predictors autoregressively, by sampling one target and appending it to the features. This paper is the first to evaluate whether this setup induces a valid stochastic process. Informally, we ask: *Do we get the same answer to the same question asked in two different ways?* The standard Kolmogorov consistency conditions cannot be checked on TFMs directly, since TFMs are unable to model certain conditional distributions. We therefore propose two adapted, testable requirements. Marginalization consistency demands that marginalized conditionals equal directly predicted marginals, and factorization consistency that both factorization orders yield the same joint. Since the data-generating distribution satisfies both, a perfect TFM would as well, and every violation lower-bounds how far the model's predictions are from the ground truth. Every TFM we evaluate violates both requirements, for classification and regression, on all datasets.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06280

DyPSI: Dynamic Physics Sensing via Joint Field and Sensor-Trajectory Generation

Yizhou Zhang, Panqi Chen, Lei Cheng, Ting Zhang, Jianlong Li, Jianfeng Shi, Shikai Fang

原文摘要

Physics sensing in evolving environments requires reconstructing dense spatiotemporal fields from sparse observations while adapting where future observations should be collected. Existing reconstruction and sensor-placement methods have made substantial progress, but they typically rely on static sensing assumptions: they either use a single layout across all frames or optimize placements independently at each frame, which limits their ability to produce temporally coherent sensing trajectories. We introduce DyPSI, a dynamic physics sensing framework that formulates this problem as the joint generation of future fields and sensor trajectories conditioned on historical observations. DyPSI lifts discrete sensor coordinates into continuous Sensor Position Fields, encodes fields and sensing configurations with a shared Functional Tucker representation, and trains a joint diffusion model to generate future fields and sensing trajectories simultaneously in the resulting latent space. A Gaussian-process temporal kernel correlates the perturbation injected during diffusion training, biasing the denoiser toward temporally smooth recoveries, and we further provide an analysis linking the EDM training objective to a sequence-aggregated A-optimality criterion and a kernel-induced bound on discrete trajectory variation. Experiments on turbulent flow, GLORYS12 sea-surface temperature, and 3D car aerodynamics show that DyPSI consistently outperforms static and frame-wise placement strategies, with substantial error reduction under tight sensor budgets. The code is available at https://github.com/mengqi132/DyPSI.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2505.06896

RCOMPSs: A Scalable Runtime System for R Code Execution on Manycore Systems

Xiran Zhang, Javier Conejero, Sameh Abdulah, Jorge Ejarque, Ying Sun, Rosa M. Badia, David E. Keyes, Marc G. Genton

原文摘要

R has become a cornerstone of scientific and statistical computing due to its extensive package ecosystem, expressive syntax, and strong support for reproducible analysis. However, as data sizes and computational demands grow, native R parallelism support remains limited. This paper presents RCOMPSs, a scalable runtime system that enables efficient parallel execution of R applications on multicore and manycore systems. RCOMPSs adopts a dynamic, task-based programming model, allowing users to write code in a sequential style, while the runtime automatically handles asynchronous task execution, dependency tracking, and scheduling across available resources. We present RCOMPSs using three representative data analysis algorithms, i.e., K-nearest neighbors (KNN) classification, K-means clustering, and linear regression and evaluate their performance on two modern HPC systems: KAUST Shaheen-III and Barcelona Supercomputing Center (BSC) MareNostrum 5. Experimental results reveal that RCOMPSs demonstrates both strong and weak scalability on up to 128 cores per node and across 32 nodes. For KNN and K-means, parallel efficiency remains above 70\% in most settings, while linear regression maintains acceptable performance under shared and distributed memory configurations despite its deeper task dependencies. Overall, RCOMPSs significantly enhances the parallel capabilities of R with minimal, automated, and runtime-aware user intervention, making it a practical solution for large-scale data analytics in high-performance environments.

cs.DC · stat.CO

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2603.08462

Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck

Fabio Valerio Massoli, Andrey Kuzmin, Arash Behboodi

原文摘要

Chain-of-Thought (CoT) prompting improves LLM accuracy on complex tasks but often increases token usage and inference cost. Existing ``Budget Forcing'' methods reduce cost via fine-tuning with heuristic length penalties, suppressing both essential reasoning and redundant filler. We recast efficient reasoning as a lossy compression problem under the Information Bottleneck (IB) principle, and identify a key theoretical gap when applying naive IB to transformers: attention violates the Markov property between prompt, reasoning trace, and response. To resolve this issue, we model CoT generation under the Conditional Information Bottleneck (CIB) principle, where the reasoning trace $Z$ acts as a computational bridge that contains only the information about the response $Y$ that is not directly accessible from the prompt $X$. This yields a general Reinforcement Learning objective: maximize task reward while compressing completions under a prior over reasoning traces, subsuming common heuristics (e.g., length penalties) as special cases (e.g., uniform priors). In contrast to naive token-counting approaches, we introduce a semantic prior that measures token cost by surprisal under a language model. Crucially, the prior is queried only for token-level log-probabilities, adding negligible overhead to the training loop. Empirically, our CIB objective prunes reasoning redundancy while preserving fluency and logic, improving accuracy at moderate compression and enabling aggressive compression with minimal accuracy drop. These gains generalize across model families and task domains, confirming CIB as a domain-agnostic CoT compression framework.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06272

Evaluating Human-LLM Alignment on Visualization Preference

Tanmay Bakshi, Chenglong Wang, Simra Shahid

原文摘要

There are often multiple valid visualizations, spanning different chart types, visual encodings, and aesthetic choices, that can all support a user's goal of analyzing a dataset. Determining which of these candidate visualizations is most appropriate, however, remains challenging. Prior work has leveraged expert-curated taxonomies to guide visualization design and subsequent quality evaluation, a practical evaluation model must also account for human-centric factors. Latest visualization tools integrate Large Language Models (LLMs) to generate and refine visualizations, using LLM-as-a-judge to provide design feedback. Without a deeper understanding of whether, and in what ways, human and LLM perceptions of visualizations align, there is a real risk of deploying low-quality agents that do not reflect human judgment. We study this question through a pilot experiment - we formulate the problem as a preference task over curated chart pairs and compare the judgments of state-of-the-art vision models with those of a panel of human annotators. We establish a cautionary tale: such judges risk filtering out informative, compelling displays. We find that frontier models collapse human subjectivity, a rich signal that reveals contrasting design choices, into near-deterministic verdicts.

cs.HC

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06271

Consistency Learning and Distillation for One-Step MIMO-OFDM Channel Estimation

Fredrik Nguyen, Qiwen Wang

原文摘要

Iterative neural channel estimators can improve recovery under pilot contamination, but repeated network evaluations increase inference cost. This paper develops observation-anchored consistency and Consistency Mid-Training (CMT) estimators for one-step channel estimation in multiple-input multiple-output orthogonal frequency-division multiplexing (MIMO-OFDM) systems. Both estimators refine pilot-derived least-squares channel estimates in the delay-beam domain. Consistency learning couples endpoint predictions along noisy-to-clean paths. CMT then refines a compact consistency-trained student to predict a flow-matching teacher's endpoint from intermediate trajectory states. Across the evaluated SNR range, the compact CMT student reduces uncoded symbol error rate (SER) by 9.6-20.2% relative to the one-step flow-matching baseline (0.44-0.98 dB in logarithmic SER reduction), while using 74.5% fewer parameters and 73.0% fewer network multiply-accumulates.

cs.LG · eess.SP

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06270

NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation

Xinhao Lou, Dongye Xie, Peng Sun, Chenyi Zhuang, Kaicheng Yu, Tao Lin

原文摘要

Adversarial training enables one- and few-step image and video generation, but conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator. We present NEXTGAN, which trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces. For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update. The same objective adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories. On ImageNet $256 \times 256$, NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36. For text-to-video generation, a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps. At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively. Code is available at https://github.com/LINs-lab/NextGAN.git.

cs.CV · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2606.06179

Diffusion Models Observe Only Gradients: A Geometric Perspective on Score Matching Errors

Naïl B. Khelifa, Richard E. Turner, Ramji Venkataramanan

原文摘要

Score-based diffusion models are typically trained by minimizing the $L^2$ score matching error, and standard theoretical analyses rely on this quantity to bound the sampling discrepancy between the learned and target distributions. Recent work has shown that the score admits an orthogonal decomposition into gradient and solenoidal components in which only the gradient part matters for exact sampling. We apply this decomposition to the score error of trained models and study its quantitative consequences. Since only the gradient component of the error enters the marginal Fokker-Planck dynamics, a model can incur arbitrarily large $L^2$ score error while perfectly matching the target. Starting from this observation, we present three results. First, we derive an upper bound on the Kullback-Leibler divergence that depends only on the observable gradient component of the error. It tightens the standard Girsanov bound for generic score networks and identifies that bound's looseness as the cost of working with path-space rather than marginal-space dynamics. Second, we prove that no non-trivial function of the $L^2$ score error can uniformly lower bound any divergence between the learned and target distributions. Third, we give an estimator of the gradient component, computable from denoising score matching residuals via a dual Sobolev identity, which empirically correlates substantially better with sample quality than the full $L^2$ error.

stat.ML · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2505.04311

How the Misuse of a Dataset Harmed Semantic Clone Detection

Jens Krinke, Chaiyong Ragkhitwetsagul

原文摘要

BigCloneBench is widely used to evaluate traditional clone detection and, more recently, machine learning approaches to semantic clone detection and functional similarity. This paper demonstrates that BigCloneBench's Weakly Type-3/Type-4 (WT3/T4) subset is problematic as ground truth for learning or evaluating functional similarity. Its tagged-tagged WT3/T4 clone pairs combine two methods labelled as true positives for the same functionality and do not contain exemplar functions. A population-weighted analysis of a stratified probability sample of 406 such pairs estimated that 93.9% of pairs in the eligible tagged-tagged WT3/T4 population are not functionally similar under our operational definition and are therefore mislabelled (conservative 95% confidence interval: 91.5%-96.2%). From a literature review of 179 papers, we estimate that 128 used BigCloneBench to evaluate semantic clone detection in a way that their results could be affected by the mislabelling. High F1 scores on BigCloneBench do not establish whether an approach has learned functional similarity or relied on dataset-specific shortcuts. Our study does not assess BigCloneBench's validity for evaluating syntactic or textual clone detection of Type-1, Type-2, and Type-3 clones. We acknowledge its important contributions to traditional clone detection research. However, using its WT3/T4 subset as ground truth for same-main-function similarity without validating its labels for that task constitutes misuse. By threatening the validity of a substantial body of reported results, this practice has harmed the evidence base of semantic clone detection. We encourage more critical and responsible use of benchmarks and datasets and more rigorous validation during dataset creation and use.

cs.SE

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.40571

On the Quantum Chromatic Gap II

Lorenzo Ciardo, Antoine Mottet

原文摘要

We confirm a conjecture of the first author [SODA'26] on a strong form of pseudo-telepathy for Khot's $d$-to-$1$ games: for every $d\geq 4$, every $ε>0$, and every $k\in\mathbb N$, there is a $d$-to-$1$ label-cover game of classical value less than $ε$ that admits a perfect quantum strategy using finite-dimensional shared entanglement with measurement contexts of radius at least $k$. As a consequence, there exist graphs whose quantum chromatic number $χ_q$ is $3$ and whose classical chromatic number $χ$ is arbitrarily large. This is the first super-exponential $χ_q$ vs. $χ$ gap, and it improves on the exponential lower bounds of Brassard, Cleve, and Tapp [Phys. Rev. Lett. '99], who produced graphs for which $χ\geq 1.1397^{χ_q}$, and Lalonde, who recently achieved $χ\geq 1.1546^{χ_q}$. To prove the conjecture, we proceed as follows. We start with a ``stretched'' version of the Mermin-Peres magic square, obtained by spacing out its equations and linking them with additional long equation wires. This operation increases the radius of the initial measurement contexts of a perfect quantum strategy for the game, while preserving classical unsatisfiability. We then apply parallel repetition to make the classical value small, and we turn the system into a pseudo-telepathic multi-layered Kneser hypergraph via a PCP construction of Dinur, Guruswami, Khot, and Regev [STOC'03]. Finally, we feed the resulting hypergraph into the reduction of Fei, Minzer, and Wang from hypergraph colouring to $4$-to-$1$ games, we establish quantum completeness of this reduction, and we show that it preserves context radius.

quant-ph · cs.CC · cs.LO · math.CO · math.OA

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06268

Beyond Identity Creation: Bilinear Sybil Cost in Identities and Time

Homayoun Maleki, Nekane Sainz, Jon Legarda, Igor Santos-Grueiro

原文摘要

Sybil defenses price the creation of an identity, yet influence goes, again and again, to active identities. We study the cost of maintaining influence. A defense has a bilinear Sybil cost with ratio $A$ if keeping $s$ identities active for $T$ periods costs at least $sT/A$ honest units: the average cost per identity-window is at least $1/A$ times an honest participant's cost. Recurrence alone does not guarantee this cost. The cost can be carried over from earlier periods, shared among many identities, or be negligible when fast hardware does the work of thousands of honest devices. We give four sufficient conditions for a defense that checks every identity each period to achieve bilinear Sybil cost: identity binding, freshness, a shared deadline, and sequential, non-batchable work. They yield two guarantees: a bilinear total cost, with $A$ the speed advantage of the adversary's fastest hardware, and, because the work must be fresh and finished by a common deadline, a bound on the identities active in a period by the processors running at that time. We prove that none of the four follows from the other three. A verifiable delay function on an identity-bound input meets all four without payments, with $A$ of 2 to 8 for CPUs and about 28 for a hypothetical ASIC, against $10^3$ to $10^7$ for hash puzzles. We measure one active identity per core per period against about 80,000 under a signed heartbeat, no gain from batching up to 1,000 identities, and rejection of all tested attacks.

cs.CR

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.23931

Welfarist Control Design - How to fulfill the societal mandate in multi-agent control?

Sophie Hall, Kai Zhang, Ilia Shilov, Heinrich H. Nax, Saverio Bolognani

原文摘要

At the core of most socio-technical systems lies a scarce resource that is allocated among agents: highway lanes, public transit, road space, water rights, energy access, grid capacity, user attention, pollution rights, etc. With further automation of the underlying allocation processes, control engineers are increasingly tasked with making decisive assumptions regarding what society wants. In practice to date, design choices are largely driven by industry norms and conventions rather than a result of conscientiously responsible and ethical design. In this paper, we look at tools available to control engineers to design systems in a more principled manner in order to match the societal mandate. We consider three control design paradigms: online feedback optimization, control of Markov decision processes, and model predictive control. Beginning with aggregating individual agents' preferences into control design objectives, subsequently ensuring and certifying the fulfillment of those specifications, we argue that the feedback nature of control systems enables appropriate allocation of the shared resources in ways hitherto unparalleled.

eess.SY · cs.MA · math.OC

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06262

SonoDyno: A Sonic Dynamometer for Engine Speed and Load Estimation from Audio

Robin Doerfler, Beat Rossmy

原文摘要

Controllable engine sound synthesis, Active Sound Design and engine order analysis all build on the relationship between the sound of an engine and its momentary operating state: its speed and its load. Recordings of engine sound are abundant in archives, sample packs and online video, yet they rarely carry time-aligned annotations of that state, which limits their applicability to controllable engine sound generation. We present SonoDyno, which estimates the engine speed, the cylinder count and an engine-normalized load intensity from a microphone signal alone, with no tachometer, no bus access and no prior knowledge of the engine configuration. A convolutional scorer of 124,302 parameters reads the harmonic order energy of each frame at three time-frequency resolutions to jointly score cylinder count and speed, and a decoder imposes the rotational dynamics of the engine on the speed trajectory. A second head estimates the load intensity, the position of the torque between the engine's drag floor at 0 and its full-load ceiling at 1, from the order levels at the decoded speed. On recordings of engines it has never heard, with microphone position, engine type and cylinder count unknown, SonoDyno places 93 to 97% of the frames within 2% of the true speed, identifies the cylinder count with 80% accuracy and confuses the octave on fewer than 2% of the frames. On an unseen engine of a known type, the load intensity has a mean absolute error of 0.12, and the sign of the torque, driving or braking, is right on 91% of the frames. General pitch trackers fall behind even when the harmonic order of each engine is supplied: on our validation recordings and at a tolerance of 5%, the best of them places 62% of the frames within the tolerance, against 96% for SonoDyno.

cs.SD · eess.AS · eess.SP

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2609.34378

Marathoner: Ultra-Long-Horizon Autonomous Intelligence

Ruiyang Zhang, Jinpeng Ou, Yifan Xie, Jingang Zhou, Lirui Pan, Qingpei Guo, Zhedong Zheng

原文摘要

Humans naturally possess the ability to work persistently toward long-term goals. Given a challenging task, humans can continuously work for months or even years to accomplish a specific objective. In this paper, we propose Marathoner, an autonomous agentic model possessing the ability of ultra-long-horizon execution. Specifically, we propose a comprehensive post-training pipeline to instill this critical capability into base model. For Ultra-Long-Horizon Task Synthesis, we leverage major release PRs containing 1000+ lines of new code from diverse GitHub repositories as the primary source for synthesizing challenging task-level data. Additionally, we introduce Multi-Task Chaining, which chains multiple generated tasks into a single more challenging task, enabling the synthesis of tasks with frontier-level difficulty. For rejection sampling finetuning, we combine strong teacher model with diverse harnesses to generate trajectories on our synthesized tasks and conduct supervised finetuning on base model with rejection sampled trajectories. For reinforcement learning, cold-started model performs real-world execution through harnesses in independent sandboxes during rollout process, effectively facilitating the acquisition of genuine ultra-long-horizon execution capability. We further propose a novel reward strategy, Later Stage Bonus Reward, which explicitly encourages model to perform meaningful maneuvers during later stages of execution. Through extensive evaluation on 5 benchmarks containing ultra-long-horizon tasks, Marathoner achieves consistent and substantial performance improvements over base model and even surpasses performance of strong proprietary model. Further analysis shows that Marathoner can consistently work for 10+ hours and conduct 1000+ tool calls on highly challenging tasks.

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2610.31513

Closing All-Exp Gaps in Local Search for the Traveling Salesman Problem and Maximum Constraint Assignment

Sophia Heimann, Hung P. Hoang, Stefan Hougardy

原文摘要

We prove the all-exp property for the Traveling Salesman Problem with the $2$-opt neighborhood, and for $(3,2,2)$- and $(2,3,2)$-Maximum Constraint Assignment with the flip neighborhood. The all-exp property states that infinitely many instances have an initial solution from which every sequence of improving moves to a local optimum, and hence every run of local search, has exponential length. For the Traveling Salesman Problem, the all-exp property was previously known for every $k$-opt neighborhood with $k \ge 3$, so our result for $2$-opt completes the picture for all $k$-kopt neighborhoods. In $(p,q,r)$-Maximum Constraint Assignment, each constraint involves at most $p$ variables, each variable occurs in at most $q$ constraints, and each domain has size $r$. Decreasing any of the parameters of $(3,2,2)$ or $(2,3,2)$ yields a problem in which every sequence of improving flips has polynomial length, so both Maximum Constraint Assignment results are tight. All three results settle long-standing open problems. Our proofs build on a common construction idea, which reveals a close connection between the local-search dynamics of the three problems.

cs.DS · math.CO

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2512.15282

A Graph-Based Conceptual Modeling Framework to Design for Human-Robot Coordination

Martijn IJtsma, Salvatore Hargis

原文摘要

Studies of human-robot interaction in dynamic and unstructured environments show that as more advanced robotic capabilities are deployed, the need for cooperative competencies to support collaboration with human problem-holders increases. Designing human-robot systems to meet these demands requires an explicit understanding of the work functions and constraints that shape the feasibility of alternative joint work strategies. Yet existing human-robot interaction frameworks either emphasize computational support for real-time execution or rely on static representations for design, offering limited support for reasoning about coordination dynamics during early-stage conceptual design. To address this gap, this article presents a novel computational framework for analyzing joint work strategies in human-robot systems by integrating techniques from functional modeling with graph-theoretic representations. The framework characterizes collective work in terms of the relationships among system functions and the physical and informational structure of the work environment, while explicitly capturing how coordination demands evolve over time. Its use in conceptual design is demonstrated through a case study in disaster robotics, which shows how the framework can be used to support early trade-space exploration of human-robot coordination strategies and to identify cooperative competencies that support flexible management of coordination overhead. These results show how the framework makes coordination demands and their temporal evolution explicit, supporting design-time reasoning about cooperative competency requirements and work demands prior to implementation.

cs.RO · cs.HC

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06254

HarnessFlow: MCTS-Guided Harness Optimization for LLM Agents

Zhi Hong, Qianlong Wang, Hongchen Wei, Aidan Wu, Daohong Ying, Mingze Kong, Bei Liu, Chong Luo

原文摘要

The performance of large language model (LLM) agents depends on both the underlying model and the harness that governs interaction with tools and the environment. Yet harness design remains largely manual, and automating it requires navigating an open-ended search space through costly end-to-end evaluations. We introduce HarnessFlow, a framework for optimizing complete executable harnesses under a limited candidate-evaluation budget, with the model and task environment held fixed. HarnessFlow organizes candidates and their evaluation histories in a refinement tree, using a Monte Carlo Tree Search (MCTS)-based policy to balance exploitation and exploration. A coding agent refines selected harnesses using archived code, execution feedback, and a library of reusable harness mechanisms. With 15 refinement attempts per benchmark, HarnessFlow achieves the highest mean test scores among the compared methods on six benchmarks spanning agentic tasks, coding, and web interaction. The optimized harnesses achieve an average score of 69.48%, outperforming the initial harnesses and Meta-Harness by 10.89 and 3.95 percentage points, respectively, while incurring lower average per-task execution costs than hand-engineered baselines. These results suggest that structured harness search offers a promising route toward automating the design of capable and efficient LLM agents.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v3arXiv:2603.01891

SEAR: Sample Efficient Action Chunking Reinforcement Learning

C. F. Maximilian Nagy, Onur Celik, Emiliyan Gospodinov, Florian Seligmann, Weiran Liao, Aryan Kaushik, Gerhard Neumann

原文摘要

Action chunking improves exploration and accelerates value propagation in long-horizon reinforcement learning, but naively applying off-policy methods to the temporally extended action space at reduced decision frequency offsets these gains, leading to poor sample efficiency. Existing online action chunking methods address these issues through computationally expensive critic-only approaches or by relying on offline data. We introduce SEAR, a sample-efficient off-policy algorithm that enables online reinforcement learning with action chunks by addressing both challenges. To handle high-dimensional action sequences, SEAR employs a causal transformer critic trained with multi-horizon targets that provide a training signal for every prefix of an action chunk, effectively increasing the useful gradients per sample. To restore decision frequency, SEAR operates with a receding horizon and random replanning, ensuring uniform state coverage while combining the fast value propagation of large chunks with the reactivity of small ones. SEAR outperforms state-of-the-art online reinforcement learning methods including SimbaV2 on challenging Metaworld manipulation tasks. Beyond online RL, applying SEAR to the offline-to-online method QC improves its performance on the demanding OGBench cube-triple tasks.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06250

A Polynomial Erdős-Pósa bound for Eulerian digraphs

Esther Galby, Roohani Sharma, Bart van der Steenhoven

原文摘要

In [Combinatorica 1996], Reed, Robertson, Seymour and Thomas proved Younger's conjecture dating back to 1973 that states that directed cycles have the Erdős-Pósa property. Specifically, they showed that every directed graph $D$ contains either $k$ arc-disjoint cycles or a set of at most $f(k)$ arcs that intersect every directed cycle. The function $f$, depending only on $k$, is however notoriously large--in fact, not even elementary. The first improvement on this function was only very recently provided by Hatzel, Kreutzer, Milani, and Muzi [FOCS 2024] who showed that $f$ can be taken to be an exponential tower of height $8$. In this work, we improve this function to $O(k^4)$ in the case where the digraph is Eulerian. This also answers an open question by Wahlström [Springer Monographs in Mathematics 2018] who asked explicitly when Eulerian digraphs have a polynomial Erdős-Pósa property. The only other known result about Erdős-Pósa on Eulerian digraphs is a result from Seymour [Combinatorica 1996] who showed that $f(k)=k$ when the Eulerian digraph admits a linkless embedding in 3-space, and such an equality is not possible for general Eulerian digraphs. We also complement our positive result, by showing that the result of Seymour cannot be extended to give a linear bound ($O(k)$) on general Eulerian digraphs, by showing that even on Eulerian digraphs the lower bound for the Erdős-Pósa property is $Ω(k \log k)$.

cs.DM · cs.DS · math.CO

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06247

CAPE-JEPA - Cross-modal Adversarial Probabilistic Embedding JEPA

Aniket Junghare, Biplab Banerjee

原文摘要

Cross-modal remote sensing image retrieval (RS-CMIR) requires representations that are simultaneously modality-invariant and semantically discriminative, despite the profound physical gap between heterogeneous sensors such as Synthetic Aperture Radar (SAR) and multispectral imagery. Recent JEPA-based Predictive self-supervised approaches such as X-JEPA formulate cross-modal alignment as a deterministic latent forecasting task, implicitly assuming uniform patch predictability and leaving cross-modal distributional alignment unaddressed. We propose CAPE-JEPA, a probabilistic cross-modal Joint Embedding Predictive Architecture with three contributions: (i) a cross-attentional Gaussian predictor outputting per-patch parameters (mu, sigma^2) conditioned on opposite-modality context, trained via negative log-likelihood to model inherent cross-sensor prediction uncertainty; (ii) adversarial distributional regularization combining unimodal isotropic Gaussian regularization with a novel cross-modal adversarial variant that aligns modality distributions via worst-case Epps-Pulley characteristic function tests without explicit sample pairing; and (iii) in-training contrastive learning with attention-pooled heads for scene-level discriminative grounding. Evaluated on BigEarthNet-14K (BEN-14K) and fMoW across SAR, multispectral, and RGB modalities, CAPE-JEPA surpasses X-JEPA by up to 13.74 points on BEN-14K and 15.59 points on fMoW, outperforming all baselines including SkySense and CROMA across both unimodal and cross-modal directions while using fewer parameters (407.30M vs. 432.72M).

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.16857

Learning to Learn from Multimodal Experience

Xingyu Sui, Weixiang Zhao, Yongxin Tang, Yanyan Zhao, Yang Wu, Dandan Tu, Bing Qin

原文摘要

Multimodal agents increasingly interact with real-world-like environments and generate rich trajectories spanning perception, reasoning, and action. A key research focus is how to transform these raw multimodal trajectories into reusable experience that can improve future task performance. Existing experience-driven learning methods are mostly developed in textual settings and typically rely on manually designed memory schemas, making them less effective for multimodal experience that requires composing heterogeneous signals across modalities, temporal scales, and abstraction levels. We propose AutoMMemo, a general framework that enables agents to automatically design memory mechanisms for learning from multimodal experience. AutoMMemo represents each memory mechanism as an executable memo program, which defines how past episodes are stored, organized, and retrieved. It discovers effective memo programs through an iterative process of update-then-retrieve evaluation, reflection-guided mutation, and budget-aware tree search. Experiments across GUI/Web navigation and multimodal visual reasoning benchmarks show that AutoMMemo consistently improves agent performance over no-memory agents and fixed memory baselines, while maintaining reasonable token and interaction costs. Transfer and search-process analyses further demonstrate that the learned memory designs capture reusable experience structures and progressively refine retrieval relevance. These results highlight adaptive memory design as a key optimization target for multimodal experience-driven learning. Code is available at https://github.com/XingYuSSS/AutoMMemo.

cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.30792

Trace2World: Reconstructing App Worlds from Real Interaction Traces for Scalable GUI Agent Training

Zezhou Wang, Hanzhang Zhou, Panrong Tong, Quyu Kong, Pengxiang Li, Steven Hoi, Zhuzhong Qian, Yue Wang

原文摘要

Scaling GUI reinforcement learning requires environments that preserve real-app behavior. Yet coupling behavior inference with coding can turn partial traces into runnable mocks with incorrect state transitions. We introduce \textbf{Trace2World}, which separates global behavior reconstruction from implementation. It reconciles traces into a shared, source-linked app model, then compiles an executable environment. Independent replay of source workflows localizes mismatches for bounded implementation repair. Validated worlds then support learner-adaptive task and environment augmentation within source-supported behavior. From 1,570 traces, Trace2World constructs 50 environments. On 200 workflows across ten apps, 91.1% are reproduced end to end, versus 59.8% for the strongest matched task-local baseline. Source-grounded validation retains 94.0% of simulated training gains on real devices, versus 54.7% under self-validation. Learner guidance raises benchmark success from 56.9% to 61.1%. Training in Trace2World further improves the model by 20.2 points on our benchmark, 5.4--16.2 points across four external GUI benchmarks, and 16.3 points on 100 real-device tasks. Reliable reconstruction separates what behavior source evidence supports from whether the implementation preserves it.

cs.LG · cs.SE

查看论文
arXiv ↗PDF ↗
版本更新 · v14arXiv:2203.05022

A proof of P != NP (New symmetric encryption algorithm against any linear attacks and differential attacks)

Gao Ming

原文摘要

P vs NP problem is the most important unresolved problem in the field of computational complexity. Its impact has penetrated into all aspects of algorithm design, especially in the field of cryptography. The security of cryptographic algorithms based on short keys depends on whether P is equal to NP. In fact, the security requirements for cryptographic keys are much stricter than those for P$\neq$NP, the security of the key must ensure not only a sufficiently high computational complexity to crack it, but also consider the security of each bit of the key, while fully avoiding the effectiveness of various attack methods. In this paper, we innovatively propose a new encoding mechanism and develop a novel block symmetric encryption algorithm, whose encryption and decryption can be completed in linear time. The key consists of 12 variables, for the attacker, in the case when only the plaintext-ciphertext correspondence is known, the problem of cracking the key is equivalent to solving a system of equations about 12 unknown variables $(w_{1}, w_{2}, ..., w_{11}, w_{12})$ together with some implicit variables which are generated by the encryption process. In this paper, we will prove that there is no universal algorithm without brute-force searching at least one variable during the process of verifying five variables $(w_{1}$, $w_{2}$, $w_{3}$, $w_{4}$, $w_{9})$ whose computational complexity is not greater than solving the 12 variables......

cs.CC · cs.IT

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06244

Rubric Curriculum RL: Exploiting the Generation-Verification Gap in Non-Verifiable Domains

Sumeet Ramesh Motwani, Tejas Krishnan, Alistair Letcher, Charles London, Suhaas Bhat, Huitian Jiao, Philip Torr, Riashat Islam 等 13 位作者

原文摘要

Reinforcement learning with verifiable rewards (RLVR) has led to significant gains in math and code generation. Extending these gains to open-ended domains remains challenging: ground-truth verification is unavailable, human annotation is expensive, and learned reward models are prone to reward hacking. The generation-verification gap, where judging good outputs is easier than producing them, offers a potential path to improvement. However, rubric-based judges are often imprecise and stationary, so policies can overfit or reward hack them. We introduce Rubric Curriculum RL (RcRL), which exploits this gap without new data, human annotations, or stronger teacher models. We combine pairwise preferences over rollouts with a curriculum over rubric criteria, yielding a more discriminative signal than absolute scoring and a non-stationary objective. Whereas baselines plateau within a few hundred steps, RcRL preserves output entropy and keeps improving past 1000 steps. On creative writing, RcRL achieves a 70.5% human win rate against the base model, with gains across benchmarks and judges. On HealthBench, RcRL improves over the base model by 14.6% (full set) and 25.4% (consensus subset), outperforming all training baselines, including instance-specific rubrics.

cs.LG · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2611.05206

Diffusion Plug-in Priors for Self-Supervised Ultrasound Video Restoration

Chu Chen, Mingduo He, Yining Li, Yinhao Wu, Wei Tang, Yiwen Luo, Yixuan Chen, Pasquale Cascarano 等 9 位作者

原文摘要

Ultrasound videos are limited by hardware-constrained spatial resolution and signal-dependent speckle, while paired low- and high-resolution acquisitions for supervised restoration are rarely available. We aim to jointly super-resolve and despeckle ultrasound videos without paired training data or network retraining. We propose UltraDiff, which uses a frozen pre-trained diffusion model (DM) as a plug-in prior and restores each frame by optimizing the latent seed of the DM sampler under a range-constrained maximum a posteriori formulation. We derive a speckle-aware data-fidelity term from a Rayleigh scale-family model and characterize its relative-residual and variance-adaptive weighting properties. We further introduce a stabilized fidelity that is bounded below and strictly convex in log intensity for positive observations in its unclamped form. A total-variation regularizer complements the DM prior, and a temporal seed update warm-starts each frame from the optimized seed of its predecessor. On EchoNet-LVH videos with simulated speckle and additive noise, UltraDiff with a natural-image prior improves PSNR by 1.24\,dB over the strongest baseline, and ultrasound-related priors improve it further. On real clinical echocardiography videos, UltraDiff reduces the mean absolute error of zero-shot ejection fraction (EF) prediction from 7.31 to 6.37 EF points relative to DUP, the strongest baseline. Combining a speckle-aware fidelity with frozen diffusion priors enables ground-truth-free ultrasound video restoration that benefits downstream measurement. UltraDiff needs neither paired data nor retraining and is not tied to a specific diffusion checkpoint. Code associated with this work will be made publicly available upon acceptance.

cs.CV

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2608.15008

Harness the Memory: A Holistic Evaluation of Memory Substrates in Memory Agents

Wei-Chieh Huang, Weizhi Zhang, Yuchen Wu, Yankai Chen, Eric Hanchen Jiang, Wooseong Yang, Yiwei Yang, Henry Peng Zou 等 15 位作者

原文摘要

Memory is becoming core infrastructure for long-horizon LLM agents, yet existing evaluations offer limited guidance on which memory substrate, namely the underlying medium in which memory is represented and stored, should be used under different operating regimes. We introduce a unified suite of 26 performance and efficiency metrics, including auxiliary-LLM costs attributed to the write, read, and management stages, and apply it identically to 11 memory methods in a controlled harness that varies only the memory substrate. The methods cover dense and sparse indices, text records, structural stores, hierarchical stores, refinement-based memories, parametric updates, and activation-compatible context mechanisms, and we evaluate them with three backbone models on four benchmark suites spanning user-centric question answering and agent-centric decision-making. Our results show that no single substrate consistently dominates: broad retrieval benefits long-context factual QA, while excessive retrieval can harm sequential decision-making by shifting attention away from action-critical context. Scalability introduces a further routing axis, as substrates that perform well at moderate history lengths can become costly or brittle at longer horizons. These regime-dependent reversals make substrate selection a design axis. Choosing a substrate per observable operating regime improves over the best fixed substrate on our benchmarks, while query-level routing leaves most of the oracle headroom open. The findings provide empirical guidance for designing efficient, reliable, and regime-aware long-term memory for LLM agents.

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.06238

Zuse: A Decision Model for European Languages

Umair Mehmood Imam

原文摘要

Decision models read a document and return a probability for each of a fixed set of options instead of generating text. Several appeared in 2026, but none has published results outside English. We describe Zuse, three decision models (4B, 9B and 27B) that train a low-rank adapter and a pointer head on frozen open-weight bases, targeting 27 European languages, and use them to study how this design behaves across languages and tasks. On a broad European suite of 11 sources excluded from training, covering 9 task groups and 27 languages, a development set that also guided model selection, the 27B model reaches 87.5% accuracy with a pooled expected calibration error of 0.016, and 0.035 when weighted per source. Post-hoc analyses of these single-seed results give three findings. First, miscalibration is structured by task rather than language: on parallel reading and topic items in 26 languages the confidence gap varies about as much as sampling noise, while across sources it spans -0.07 to +0.09, with knowledge tasks overconfident and acceptability judgements underconfident at all three sizes, so one fitted temperature cannot correct both. Second, the share of decisions that can be automated at a 5% error budget follows chance-corrected skill (Spearman 0.97 over 33 source and model pairs) more closely than raw accuracy (0.91). Third, reach beyond the training languages is set by the base model: across 210 languages, changing the training run moved scores by at most about 2 points within a size, while languages outside the base's documented list trail by 23 to 28 points at every size. On English public suites Zuse is behind Jev and Kev, including Kev models on the same bases. We also report two pitfalls in building multi-source training data: label-word collision between topic schemes, and an injection augmentation that taught a shortcut. Zuse can be tried at https://bineric.com/zuse.

cs.CL · cs.LG

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2607.15215

Stochastic Ising networks with asymmetric and time-delayed interactions

Hantao Zhang, Sidra Gibeault, Matthew W. Daniels, Philippe Talatchian, Ursula Ebels, Advait Madhavan, Mark D. Stiles

原文摘要

Stochastic Ising networks are widely used to describe collective dynamics in complex systems and to compute, yet physical implementations of these networks often contain both asymmetric interactions and nonzero signal propagation times that fall outside conventional theories of their behavior. Here, we study stochastic Ising networks with asymmetric and time-delayed interactions motivated by experimental observations in coupled superparamagnetic tunnel junctions. Time delay fundamentally reshapes the dynamics induced by antisymmetric couplings, producing strong oscillatory temporal correlations consistent with experiment. Sufficiently long delays strongly suppress the non-uniformity of steady-state probability distributions, a behavior explained by an exponential suppression of the first-order equal-time cross-correlation between pairs of spins. Residual non-uniformity is due to the equal-time cross-correlation mediated by a third spin that contributes only at higher order in the coupling strength. These relatively featureless probability distributions coexist with pronounced delayed temporal correlations, distinguishing them from equilibrium high-temperature behavior. We further show analytically that these delay-induced near-uniform distributions emerge in a broad class of stochastic networks that lack symmetry-breaking bias fields. Introducing symmetry-breaking bias fields restores a strong dependence of the steady state on interactions with qualitatively modified behavior. Simulations of networks with five coupled spins demonstrate that these effects persist beyond minimal systems with only two spins.

physics.app-ph · cond-mat.mes-hall · cond-mat.stat-mech · cs.ET · physics.data-an

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2604.23334

A Lagrangian Approach to Interdiction of Linear Minimization Problems

Yu Cong, Kangyi Tian

原文摘要

We study budgeted interdiction of linear minimization problems. Given a family of feasible sets, element weights, and deletion costs, the objective is to delete elements within a prescribed budget and minimize the remaining weight of a feasible set. We give an alternative, simpler Lagrangian approach that generalizes the connectivity interdiction framework of Huang, Obscura Acosta, and Yingchareonthawornchai (IPCO' 24). When the interdiction optimum is positive, we show that the optimal Lagrange multiplier gives truncated element weights, in which every feasible set supporting an optimal interdiction solution has weight less than twice the minimum under the truncated weights. We apply the reduction to graph cuts, $k$ cuts, and cocircuits of perturbed graphic matroids. We obtain deterministic fully polynomial-time approximation schemes for graph connectivity interdiction and for cogirth interdiction on perturbed graphic matroids of fixed perturbation rank. For graph connectivity, the running time is $\tilde O(n^3m^2/\varepsilon)$. The $k$-cut application gives a fully polynomial-time randomized approximation scheme.

cs.DS

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.34218

From Presence Labels to Bounding Boxes: Can Multimodal Large Language Models Scale Surgical Instrument Localization?

Mohamed Hamdy, Muraam Abdel-Ghani, Fatmaelzahraa Ahmed, Sifna Nasar, Mariam Ahmed, Khalid Al-Jalham, Abdulaziz Al-Ali, Shidin Balakrishnan

原文摘要

Spatial localization of surgical instruments through bounding-box annotations is a foundational component of surgical scene understanding, supporting detection, tracking, interaction analysis, and grounded vision-language applications. Yet existing annotations often cover selected frames, video subsets, or dataset variants rather than complete procedures. Manual extension is time-consuming and expert-dependent. Additionally, weakly supervised and detector-based pseudo-labels require task-specific training and may be imprecise. In this study, we design a scalable MLLM-based annotation framework that converts surgical frames and existing instrument-presence labels into accurate bounding boxes, and systematically analyze its reliability, cost, and design sensitivity. We instantiate the framework using six MLLMs and a multi-call configuration on laparoscopic cholecystectomy in Cholec80. Using a carefully designed stratification scheme, two annotators independently label 1,000 frames, and an expert surgeon reviews one annotation set to establish the reference. Results show that instrument-set F1 ranges from 88.2 to 92.7,whereas mIoU ranges from 0.180 to 0.735, showing that comparable recognition can conceal large localization differences. A structured multi-stage annotation pipeline achieves 0.784 mIoU, compared with 0.735 for the best single-call baseline and 0.814 human-human agreement. Using the best accuracy pipeline, we create and publicly release SurgBox, a spatial extension of Cholec80 comprising 285,977 bounding boxes across all 184,498 annotated frames from the 80 videos. The annotations and human references are made available on GitHub

cs.CV

查看论文
arXiv ↗PDF ↗