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

+197新增论文
+151其中 AI 相关
+16已有论文修订
+95新增 AI 导读

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

统计口径与数据新鲜度

统计 2026-11-20 20:52 — 2026-11-21 20:52(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

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

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

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

浏览论文库

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

找到 6,976 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-21 12:17(北京时间)
新论文 · v1arXiv:2611.19575

Risk-Averse Decision-Focused Learning: Minimizing CVaR of Regret

Jayanta Mandi, Senne Berden, Victor Bucarey, Concepcion Dominguez, Tias Guns

原文摘要

Many combinatorial optimization (CO) problems involve parameters that are uncertain when the problem must be solved. In such settings, the parameters have to be predicted from correlated contextual information before a solution can be computed. \emph{Decision-focused learning} (DFL) addresses this by training the predictive model to produce predictions that optimize downstream decision quality. Most existing DFL methods are risk-neutral and minimize the expected regret. In high-stakes applications this can lead to undesirable outcomes, because decision-makers must also account for the tail risk of decision quality. Prior risk-sensitive approaches typically model uncertainty explicitly and solve a stochastic or robust optimization problem at inference time, which increases computational complexity. We instead propose a DFL framework that achieves risk-sensitive behavior while retaining point predictions and deterministic optimization at inference time. Starting from the dual representation of the Conditional Value-at-Risk (CVaR) and adding entropy regularization, we derive a smooth surrogate for the CVaR of regret. Its gradient induces an instance-dependent reweighting within each minibatch that emphasizes high-risk outcomes, so the approach can turn existing surrogate-loss DFL methods, including the SPO+ loss, into risk-sensitive variants. When instantiated with SPO+, the resulting objective admits a generalization bound on the CVaR of the true regret. Experiments on knapsack, portfolio, shortest-path, and energy-aware scheduling problems show substantial reductions in tail regret when the uncertain parameters exhibit heterogeneous tail behavior.

cs.LG · cs.AI

新论文 · v1arXiv:2611.19572

Spectral Steering: Local Inverse Geometry for Frozen Diffusion Generators

Angelo Moroncelli, Alessandro Leanza, Loris Roveda

原文摘要

A frozen diffusion generator may need to change several output properties while preserving others. We study when guidance faithfully realizes such a structured request and whether local inverse correction improves its direction. Our account links this benefit to the interaction between the request and the objective-specific remaining-sampler Jacobian: conditioning alone is insufficient. Spectral steering combines a request-specific cosine-gain predictor with damped inverse correction and fixed-time relinearization for finite edits. Scalar nulls and matched-movement controls test the predicted mechanism. Across 8,923 instances in a retrospectively identified local regime, predicted and realized gains correlate at Spearman 0.79, versus -0.11 for condition number. Experiments span images, audio, robotics, and control and planning, including complex character motion generation. On 50 paired inpainting cases, four corrections improve unobserved-pixel PSNR over Adam by 3.25 dB at equal total derivative traversals. Together, the analysis and controls connect request-dependent geometry to directional gains and establish where these gains translate into downstream performance.

cs.LG · cs.AI · cs.CV · cs.RO

版本更新 · v3arXiv:2312.10181

Integrating Fairness and Model Pruning Through Bi-level Optimization

Yucong Dai, Gen Li, Feng Luo, Xiaolong Ma, Yongkai Wu

原文摘要

Deep neural networks achieve high predictive performance but often require substantial computational and storage resources, which motivates model compression techniques such as neural network pruning. Conventional pruning methods, however, can introduce or amplify algorithmic bias and produce disparate outcomes across demographic groups. We study fair model pruning, which seeks sparse models that satisfy fairness criteria while maintaining predictive performance. We characterize fairness using both group accuracy disparity and differences in compression-induced accuracy degradation across demographic groups. We propose a framework that jointly optimizes pruning masks and model weights under fairness constraints, thereby integrating accuracy and fairness into the compression process. We formulate fair pruning as a constrained bi-level optimization problem and derive an efficient optimization procedure. Experiments across multiple datasets, predictive tasks, model architectures, and sparsity levels show that our framework improves the trade-off among fairness, accuracy, and training efficiency relative to the baseline pruning strategies. Compared with the strongest evaluated baselines, BiFP maintains competitive accuracy and fairness while requiring more than 94% fewer training iterations in the real-world experiments.

cs.LG · cs.AI · cs.CY

新论文 · v1arXiv:2611.19569

Learning from Mutants: Teaching Small Language Models to Synthesise Loop Invariants

Daragh King, Vasileios Koutavas

原文摘要

Large language models (LLMs) have proven promising for synthesising loop invariants, a long-standing obstacle to automated program verification. Yet, frontier LLMs are typically accessible only through third-party APIs, hampering reproducibility and local deployment. Small language models (SLMs) sidestep these obstacles but require task-specific fine-tuning, for which high-quality training data is needed. We present a failure-driven data curation method for fine-tuning language models to synthesise loop invariants. Inspired by mutation testing, our approach applies semantic mutation operators to ground-truth invariants, producing semantic mutants that induce verification failure. Each mutant is paired with verifier feedback and its ground-truth counterpart, yielding training examples for feedback-guided invariant refinement. Whereas the vast space of mutants is a long-standing burden in mutation testing, it becomes an asset when generating training data, multiplying the examples obtainable from a fixed set of verified programs. We implement our approach for Dafny, augmenting an existing synthetic dataset of verified programs, and introduce a framework that iteratively refines candidate invariants using verifier feedback. On 472 DafnyBench programs, our best-performing fine-tuned Gemma-4-E4B configuration verifies 50.8% of programs zero-shot, compared to 8.1% for the base model, a 6.3$\times$ improvement. Higher-order mutants prove more effective as training data than first-order mutants, and adding mutation-based examples to direct synthesis examples, in which invariants are removed rather than mutated, improves performance under iterative refinement. Our approach also transfers to Qwen3.5-4B, whose zero-shot verification rate rises from 9.3% to 53.6%.

cs.SE · cs.AI

版本更新 · v2arXiv:2605.17624

Multi-task learning on partially labeled datasets via invariant/equivariant semi-supervised learning

Miquel Martí i Rabadán, Alessandro Pieropan, Hossein Azizpour, Atsuto Maki

原文摘要

We investigate the potential of invariant and equivariant semi-supervised learning for addressing the challenges of training multi-task models on partially labeled datasets with differently structured output tasks. Specifically, we use the popular FixMatch method for invariant semi-supervised learning and its equivariant extension Dense FixMatch. We evaluate their performance on the Cityscapes and BDD100K datasets in the context of the prevalent object detection and semantic segmentation tasks in computer vision. We consider varying sizes of the subsets annotated for each task and different overlaps among them. Our results for both invariant and equivariant semi-supervised learning outperform supervised baselines in most situations, with the most significant improvements observed when fewer labeled samples are available for a task and generally better results for the latter approach. Our study suggests that invariant/equivariant learning is a promising general direction for multi-task learning from limited labeled data.

cs.CV · cs.AI · cs.LG

新论文 · v1arXiv:2611.19564

Behavioral Assurance for Socially Deployed AI Agents

Nicole H Brennan, Jiaee Cheong

原文摘要

AI agents are increasingly deployed in socially consequential settings, yet existing evaluations provide limited assurance that they behave consistently with their intended characteristics across diverse interactions. We introduce an evaluation framework for measuring the behavioral consistency of socially deployed AI agents. The framework operationalizes agent behavior along five measurable axes and uses a standardized evaluation battery spanning six socially salient interaction settings to assess behavioral differentiation, consistency, and drift. Controlled persona configurations serve as experimental anchors, enabling systematic testing of whether intended behavioral differences persist across contexts rather than appearing only as stylistic variation. Preliminary analyses demonstrate measurable differentiation across configurations and suggest associations between behavioral profiles and interaction outcomes. By shifting the focus from persona design to behavioral assurance, this work provides a foundation for evaluating whether socially deployed AI agents behave as intended, consistently and measurably, across the situations in which users encounter them.

cs.AI · cs.ET · cs.HC · cs.MA

新论文 · v1arXiv:2611.19561

Read, Then Shade? How Visual Evidence Changes Incentive-Driven Misreporting in Vision-Language Advisors

Hamidreza Hasani Balyani

原文摘要

Vision-language models (VLMs) increasingly read a scene and report it to someone who acts on the report, sometimes under objectives that differ from that person's. We ask whether a stated incentive moves what a VLM reports, and whether this changes when the value must be read from pixels rather than given as text. In seeded synthetic scenes, a model reads the position of a target on a labelled axis and reports one integer to a fixed receiver; it is paid, in simulation, for accuracy or for a report a few units to one side, and every scene is also presented with the true value as text. Across a pilot and three follow-up studies with hash-frozen protocols, four small open VLMs from three families, and 15,170 held-out trials, a stated payoff moved reports in both modalities, and exact compliance at three units was significantly lower with the image than with text for three of four models; Qwen2.5-VL 7B showed no reliable difference. Adding the true value as text to the image prompt removed the gap in every model. For InternVL3 the gap vanished on scenes it read correctly when it wrote its reading first; for Gemma 3 it persisted. For Qwen2.5-VL 3B, which read every scene correctly, the gap held in four of five payoff paraphrases, nearly vanished when the model wrote its reading before its report or when the prompt stated that the client wants the truth, vanished with an unpaid goal, and persisted when the model reported before stating its reading; how often it shaded at all, even with text, depended strongly on the wording. When told that the client wants the true value, three of four models still shaded every text report. Visual input did not make these advisors reliably more truthful; it made compliance with a stated incentive depend on whether the read value is made explicit and on output order, a dependence that text-only evaluations and mean-shift metrics do not reveal.

cs.LG · cs.AI · cs.CV

新论文 · v1arXiv:2611.19560

SemMask: Semantic-Prioritized Progressive Masking for Diffusion Language Models

Qi Chen, Zifeng Cheng, Zhiwei Jiang, Qing Gu

原文摘要

Diffusion Language Models (DLMs) are designed to enable globally parallel generation. However, in practice, they often exhibit pseudo-autoregressive behavior and a function-word-first bias, which undermines their performance on tasks such as text writing and code generation that benefit from global planning and parallel decoding. To address this, we propose SemMask, a novel fine-tuning paradigm that leverages a semantic-prioritized progressive masking strategy to encourage the model to decode high-information semantic tokens early and globally, thereby enhancing its global planning and parallel generation capabilities. Specifically, we first use the gradients of the original DLM with respect to each token to identify semantic tokens. During masking, these semantic tokens are preserved with priority, establishing a progressive generation order in which semantic tokens are decoded prior to function tokens. Meanwhile, we adjust the loss weight for semantic tokens to further reinforce their early prediction. Experimental results show that SemMask achieves quality comparable to strong autoregressive baselines on both text writing and code generation tasks, while providing superior semantic coherence. Notably, SemMask achieves higher accuracy with substantially fewer decoding steps, indicating that the early establishment of semantic structure improves the efficiency--accuracy trade-off.

cs.CL · cs.AI

新论文 · v1arXiv:2611.19554

Inquest: Log-Guided Root-Cause Code Localization with a Staged Agent

Yuvraj Sehgal, Sneh Patel, Mahsa Panahandeh, Naser Ezzati-Jivan

原文摘要

When a production issue is reported, engineers get a description and some logs, but no failing test and no way to reproduce the fault. The logs show where the failure was noticed, not where it was caused: in 49% of the real reports we study, the text names none of the files that the fix later changed. We built a benchmark for this setting from two public datasets and 27 open-source projects (Apache, Red Hat, MongoDB, MariaDB, Qt and Sakai). It holds 38,001 resolved issue reports with their pasted logs, the code as it was while the bug was still present, and the files the fix changed. We present INQUEST, an agent that finds the faulty code using a 27B open-weights model. Code, not the model, controls the procedure. It first gathers the evidence, including the files fixed in similar past issues that were resolved before the report was filed. The model then makes only three bounded decisions: which files to open, what role each file plays in the failure, and how to rank them. On 26,544 evaluation reports, INQUEST ranks a changed file first for 68.6% and in the top five for 85.9%, names the correct module for 90.2%, and ranks a changed function first for 53.8%. On the reports that name no faulty file, it reaches 52.8% and 73.6%. A free agent loop with the same model and tools finds files equally well, but names the faulty function 14 points less often. On a 1,000-report sample, INQUEST beats our reimplementations of two published agents by 7 and 20 points of top-1 accuracy (4 and 16 counting only the reports those agents answered), and the same model used as a re-ranker by 6. We report the cost of every system and the results that go against us, and we release the benchmark, both systems and every run transcript.

cs.SE · cs.AI

版本更新 · v2arXiv:2607.15898

Orbis 2: A Hierarchical World Model for Driving

Sudhanshu Mittal, Arian Mousakhan, Silvio Galesso, Karim Farid, Johannes Dienert, Rajat Sahay, Muhammad Ali, Thomas Brox

原文摘要

Current world models operate at a single level of abstraction, with most prioritizing perceptual fidelity while lacking the spatial reasoning and semantic understanding required for real-world downstream tasks. We present a hierarchical driving world model that factorizes future prediction across two levels operating at distinct temporal and abstraction scales: a high-level predictor that forecasts coarse scene structure over extended temporal horizons, and a low-level generator that produces detailed predictions conditioned on the high-level output. This decomposition yields high perceptual fidelity while also capturing strong spatial and semantic representations. We further show that pretraining with a diffusion forcing objective yields substantially richer internal representations than the standard teacher forcing objective, while teacher forcing -- predicting only the next frame from clean context -- produces more stable autoregressive rollouts. We therefore introduce a generic two-stage training paradigm that pretrains the model with diffusion forcing and fine-tunes with teacher forcing, combining the representational benefits of the former with the rollout stability of the latter. Our approach achieves state-of-the-art results across the standard suite of driving world model evaluations on established benchmarks, including long-horizon generation fidelity, steering responsiveness evaluated on counterfactual scenarios, and internal representation quality. Project page with code, demo, checkpoints and qualitative results: https://lmb-freiburg.github.io/orbis2.github.io/

cs.CV · cs.AI · cs.LG · cs.RO

新论文 · v1arXiv:2611.19550

Lesion-Aware Explainable AI for Diabetic Retinopathy Grading from Fundus Images

Truong Thinh Nguyen, Hai Phong Nguyen, Truong Thanh Hung Nguyen, Giang Son Tran, Hung Cao

原文摘要

Post-hoc explanations of deep-learning models for diabetic retinopathy (DR) grading are difficult to trust because different attribution methods highlight different retinal regions for the same image, and a visually plausible heatmap does not show that the highlighted pixels drive the prediction. We propose the Consensus-Calibrated Explanation Map (CCEM), which fuses Grad-CAM++, Ada-SISE, and SmoothGrad Integrated Gradients (SG-IG) using image-specific weights derived from annotation-free reliability cues: deletion sensitivity, inter-method agreement, compactness, retinal containment, and peak structure. CCEM explains a frozen ordinal grader, an EfficientNet-B7 with CBAM and a scalar regression head, pre-trained on EyePACS 2015 in Stage 1 and fine-tuned on APTOS 2019 in Stage 2 with quadratic weighted kappa (QWK)-optimized thresholds; it reaches a QWK of 0.920 on the hidden Kaggle test set and 0.559 on the external MAPLES-DR cohort. Explanations are evaluated on MAPLES-DR, 198 MESSIDOR images with expert pixel-level lesion annotations that are never used for training, threshold fitting, or fusion. On the 193 images with annotated lesions, CCEM obtains the highest Soft Pointing Game score (69.61%; 95% CI 63.7-75.3), defined as the ratio of the strongest lesion-supported saliency to the global map maximum rather than a lesion detection rate, together with the lowest deletion AUC (0.570) and the best deletion-insertion balance (0.115); each gain is significant against the strongest single explainer on paired tests. Annotation-free, per-image reliability weighting therefore turns three disagreeing explainers into one map whose peak evidence is more lesion-supported and more faithful to the model than that of any single explainer.

cs.CV · cs.AI

新论文 · v1arXiv:2611.19549

Beyond Prompts: Runtime Contracts for Coding Agents

Reda Heddad

原文摘要

Coding agents act on repositories through shell commands and file edits, and teams steer them with written engineering rules. Industrial settings expect such rules to be enforced and their application to be evidenced, and a rule given as prompt text provides neither. We test whether executable contracts enforced while the agent works reduce rule violations compared with the same rules in the prompt, retrieved as guidance, or checked only after generation. Violations are scored by an oracle whose methods differ from the in-loop checks, and the oracle is validated by two human reviewers and a model reviewer. Across three studies on two model families (240, 112 and 108 runs), enforcement during execution produced no oracle-detected violation in 104 paired runs, against 21 to 55% of runs without rules, and no loss of functional correctness in the two confirmatory studies. In the confirmatory study on tasks aimed at known failure situations (Study 3, protocol hashed before the first run), enforcement lowered the share of violating runs by 31 percentage points relative to no rules (Holm-adjusted p = 0.014); the 17-point difference from prompted rules was not significant (p = 0.12), and neither was any contrast in Study 2, where only 3 of 14 tasks produced a violation. Prompted rules removed some violation types but not recursive deletes of scratch directories, and checks after generation repaired file edits but not commands that had already run. That enforcement satisfies the rules it enforces is expected by construction; what the data add is where prompting falls short, the absence of a functional cost, and the limits of the rules and of the oracle themselves. The studies use low-cost models and one small fixture.

cs.SE · cs.AI

新论文 · v1arXiv:2611.19546

Evidence-Conditioned Completion and Synthesis of Electronic Health Records

Sicheng Shu, Parker Knight, Huitao Li, Yuan Luo, Yifan Peng, Shawn N. Murphy, Rui Yang, Chuan Hong

原文摘要

Patient information can guide electronic health record (EHR) generation, but how its amount shapes predictive utility and disclosure risk remains poorly characterized. Here we present EHRDiB, which unifies visit completion and synthesis by distinguishing original missingness from withheld inputs and combining a shared multimodal representation with an evidence-conditioned latent diffusion bridge. We compared EHRDiB with three completion and eight synthesis methods on MIMIC-IV and numerical and discretized Duke Health representations. Evaluation covered downstream predictive utility, generative diversity, and distributional fidelity for both tasks, and disclosure risk for synthesis. EHRDiB achieved the smallest utility gaps from real-data-trained classifiers in seven of eight completion comparisons. In synthesis, AUPRC gaps were 45.2% smaller for ten MIMIC-IV principal-diagnosis targets and 53.3% smaller for numerical Duke Health data relative to the best comparator. Synthesis with calibrated thresholding yielded the lowest marginal distribution error across all three datasets, with Coverage above 0.95 for both output rules. Across twelve evidence-fraction intervals in MIMIC-IV synthesis, membership-inference AUROC increased from 0.516 to 0.598 as the mean target evidence fraction rose from 0.20 to 0.50; utility gaps narrowed relative to the lowest evidence configuration but did not decrease monotonically. Our study provides a unified framework and empirical foundation for EHR generation that considers both its utility for clinical analysis and its information disclosure risk.

cs.AI · stat.AP

版本更新 · v4arXiv:2605.25796

SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustness

Jiahao Huo, Wenjie Qu, Yibo Yan, Kening Zheng, Jiaheng Zhang, Xuming Hu, Philip S. Yu, Mingxun Zhou

原文摘要

Semantic-level watermarking (SWM) improves robustness against text modifications by treating sentences as the basic unit. However, robustness to paragraph-level paraphrasing remains difficult because such attacks globally disrupt watermark signals by changing sentence order. In this work, we propose SAMark, a self-anchored watermarking framework that removes the dependency on sentence order by establishing a step-independent green region in semantic space. To improve detectability, we introduce a multi-channel hyperbolic scoring mechanism that amplifies watermark signals while suppressing noise from weakly aligned candidates. We further propose a diversity-aware filtering strategy that combines hard filtering with soft regularization, extending beyond simple n-gram repetition filters to address semantic redundancy. Experimental results show that SAMark achieves up to 90.2% TP@FP1% under typical paragraph-level paraphrasing attacks, outperforming the strongest prior baseline by more than 30% on average, while maintaining generation quality competitive with unwatermarked text and breaking the robustness-quality trade-off that limits prior methods. Our code will be released at [this URL](https://github.com/Z1zs/SAMark).

cs.CR · cs.AI · cs.CL

版本更新 · v2arXiv:2611.12942

Tabular Foundation Models vs Tuned Gradient Boosting for Telecom Churn: A Profit- and Calibration-Aware Evaluation

Ismail Hossain Polas

原文摘要

A churn model is useful through the customers it selects for a retention offer: the better model is the one whose selection makes the campaign more profitable. Tabular foundation models (TFMs), pretrained networks that predict from a table of labelled examples without being trained on it, were recently reported to decisively outperform tree ensembles for churn. That comparison, however, used untuned gradient-boosted trees (GBDTs) and ranking metrics only. We ask whether the advantage holds against tuned GBDTs when models are judged by expected campaign profit and by the reliability of their probabilities. Four TFMs (TabPFN-2.5, TabPFN-3.5, TabPFN-3.5-Fast, TabICL v2) face LightGBM, XGBoost and CatBoost tuned with Optuna on validation AUC, on five public telecom datasets, three from real operators, with 10 repeated splits each and 20 pre-declared comparisons tested with a corrected t-test and Holm correction. TabPFN-3.5 earns significantly more expected maximum profit (EMP) than each tuned GBDT on two datasets, Cell2Cell (1.9%) and Iranian (7.6%). On Orange it is ahead in every split (14.3%), but not significantly after correction, and no TFM differs significantly from the GBDTs on the two fictional datasets. Across 27 assumed combinations of customer value, incentive cost and acceptance rate, TabPFN-3.5 stays ahead on all three real-operator datasets, though the size of its gain varies widely. Retuning the GBDTs on EMP moved no gap by more than 0.014 per customer. Ranking and probability reliability diverge: raw TabICL v2 underpredicts churn on Orange, and a threshold policy built on its probabilities loses 15.7% of attainable profit until it is recalibrated. Because the splits resample the same customers, the results describe these datasets, not telecom operators in general. The non-commercial licence of the TabPFN weights, rather than scoring cost, restricts production use.

cs.AI · cs.LG

新论文 · v1arXiv:2611.19542

Func-JEPA: Learning a Sobolev Space of Functions as Latent Vectors

Alejandro Moreno Arcas

原文摘要

Pretraining an encoder once and reusing it across tasks is a widespread approach in vision and language. However, neural symbolic mathematics still trains a dedicated sequence-to-sequence model for each task, so that no representation of a function is shared. In this work, we propose the Function Joint-Embedding Predictive Architecture (Func-JEPA, abbreviated f-JEPA): an encoder of mathematical functions, pretrained on their symbolic form, whose latent space imitates the Sobolev space H^1 and in which differentiation is a learned latent operator. The norm of a latent is computed exactly from the function, whereas its direction is learned to match Sobolev angles. The frozen encoder is then reused by lightweight modules for decoding, composition, integration, and regression from samples. The representation and four downstream tasks are evaluated within and outside the pretraining distribution. f-JEPA recovers exact formulas from noisy samples more often than the compared symbolic-regression methods, whereas computer algebra and task-specific models remain stronger on standard integration and differential-equation benchmarks.

cs.LG · cs.AI · math.NA

新论文 · v1arXiv:2611.19537

Continuation-Guided LLM Evolution of Program Policies and Goal Compilers in Craftax

Víctor Gallego

原文摘要

Coding agents can synthesize and optimize code-based policies from simulator feedback, allowing the resulting program to act with zero inference model calls. But for long horizon tasks, two challenges arise: local improvements can hurt final success and switching to a new objective requires another expensive synthesis run. We propose a framework that first develops competence by refining the policy through a feedback cascade, moving from episode returns to hierarchical goals and belief updates which can be scored densely from simulator. In the Craftax game, a partially observed and procedurally generated environment, refinements via episode level feedback stall at <1% final-level arrival, whereas with the feedback cascade the optimized policy reaches the final stage 50% of time (28% final boss defeats). Furthermore, to make this competence reusable towards new goals, we factor the previous controller into an environment interpreter $I$ (models, memory, skills) and a rule program $p_0$, and we show that coding agents can optimize a reusable goal compiler $C_θ$ that maps structured tasks $g$ into new rule programs $p_g$. Across five independent compiler searches trained only on short tasks, the resulting compilers successfully transfer zero-shot to longer tasks without any task-specific tuning nor inference-time LLM calls: continuation-scored feedback and a fixed machine with an editable program together turn a single-task competent policy into a reusable one. Code at: https://github.com/vicgalle/craftax-program-policies

cs.AI

新论文 · v1arXiv:2611.19536

Agent Authenticator: Passkey-Grade, Hardware-Bound Authentication for AI Agents

Harish Santhanalakshmi Ganesan

原文摘要

AI coding agents now push code, publish packages, call model APIs and operate cloud resources on behalf of developers. They authenticate to these services with the same bearer secrets humans use, personal access tokens, API keys and package-registry tokens, stored where the model, its subprocesses and any co-resident malware can read them. Incidents in 2025 and 2026, including prompt-injection attacks on coding agents and self-propagating supply-chain worms, show that this arrangement fails in predictable ways. Passkeys solved the analogous problem for humans, but they require a human gesture on every use and cannot be operated by an autonomous process. We present the Agent Authenticator, a design that gives agents passkey-grade authentication properties: per-service hardware-bound keys, no shared secrets, origin binding enforced by the runtime rather than the model, normalized attestation of the execution environment, human-signed scoped mandates, and cross-device human approval for high-risk actions. The same protocol serves agents on developer laptops and in cloud enclaves, and it deploys as a profile over existing mechanisms, MCP authorization, OAuth token exchange including the Identity Assertion JWT Authorization Grant, git and SSH, rather than as a new identity system. The authenticator also produces a signed, hash-chained record of each action it authorizes. We analyze the design against a threat model centered on prompt injection and same-host compromise, map it onto five documented incidents, and discuss its limits.

cs.CR · cs.AI

新论文 · v1arXiv:2611.19530

LLM Trading Agents Earn Beta, Not Alpha: A Look-Ahead-Free Benchmark

Jackson Wang

原文摘要

Large language model (LLM) trading agents now trade stocks for their users, and benchmarks rank models by their agents' returns. Yet a single run's return says little about skill. In our study, two runs of the same model that differed only in the random seed ended 5.2 points apart on average, so telling skill from luck takes many backtests across markets, models and seeds. However, live arenas trade each day only once, and backtests either use the few months after a model's training cutoff or run on years the model may remember, often with news joined to prices by calendar date. In this paper, we present rta-bench-1, a backtest over three US markets (a rally, a bear market and a crash). It shows each headline only after its publication minute, plays each period blind and named, and checks every run for look-ahead. It also splits each return into factor exposure and alpha and scores every decision against random choices of the same day and sector. In 91 runs of six models from four vendors, no alpha was significant, market and style exposure accounted for a median 78% of the return, and the decisions did no better than random stocks of the same sector. The agents bought the stocks with the most positive news, yet in ablations they earned as much without their headlines. Still, the agents carried out rules reliably, since a single cash rule in the prompt cut a GPT-based agent's share in stocks by 30.3 points in all six pairs. Builders should therefore supply the edge themselves and report exposures and alpha, and benchmarks should replay news by publication time.

cs.AI · cs.MA

版本更新 · v2arXiv:2608.27909

Low-Altitude Fluid Antenna Network with Multi-Agent Reinforcement Learning

Tong Zhang, Yanfei Su, Shuai Wang, Wanli Ni, Chengzhong Xu, Huseyin Arslan

原文摘要

Low-altitude wireless networks (LAWNs) integrate terrestrial and aerial platforms to provide ubiquitous communication, sensing, and localization services for unmanned aerial vehicles (UAVs) and electric vertical takeoff and landing (eVTOL) aircraft. However, dynamic air-ground and air-air channels, abrupt blockages, and heterogeneous interference hinder the realization of this goal. Nevertheless, fluid antenna (FA), a cutting-edge multiple-input multiple-output (MIMO) technique, overcomes these challenges by reconfiguring antenna positions to unlock additional spatial degrees-of-freedom. In this paper, towards bringing low-altitude FA networks into reality, we study the fast and high-performance FA reconfiguration for low-altitude FA networks with multi-agent reinforcement learning (MARL). Specifically, we present an electromagnetic digital twin (EM-DT)-assisted MARL framework. To fill the sim-to-real gap, we introduce a two-stage transfer learning framework. Our case study shows that joint FA positions and beamforming optimization can enhance the system sum-rate by 118.5%, compared to the fixed position baseline. This gain comes from the dynamic millisecond timescale reconfiguration of FA arrays and the adaptive steering of beams toward aerial users with mobility.

cs.IT · cs.AI

新论文 · v1arXiv:2611.19529

Ultra-CAN: A Lightweight Blind ISAC Receiver for Chaos-Based Telemetry and Local-Seed Ranging in the Internet of Drones

Ettore Caruso, Marco Siino, Giacomo Morabito

原文摘要

The Internet of Drones (IoD) needs secure telemetry and proximity sensing within the SWaP-C constraints. We show that a single blind neural receiver driven by one chaos-based waveform performs both functions without a separate radar transceiver. In telemetry mode it decodes a peer's burst blindly; in radar mode it regenerates the per-burst chaotic reference locally and ranges the echoes of its own transmission. Ultra-CAN pairs a shared $64$-dimensional representation with two ultra-light backbones, an attention-pooled Conv1D network and a multi-head QKV network ($27.6$k--$44.2$k parameters), feeding communication and sensing heads. Both ship as int8 TensorFlow Lite Micro firmware; the Conv1D receiver decides a burst in $16.9$ ms on a $240$ MHz microcontroller with $24.2$ KiB of activations, a measured energy of $2.2$-$3.4$ mJ per burst, and float-equivalent decisions, sustaining up to $59$ bit/s of telemetry or a $7.4$ Hz range track. The two variants rank first and second among learned demodulators in four aerial swarm scenarios, with a pooled BER down to $4.6 \times 10^{-4}$ on the single-echo link ($2.4 \times 10^{-3}$ with three echoes). Its sensing head identifies true obstacles in $66.8\%$ of peer-laden bursts (vs $42\%$ matched filtering, $44\%$ OFDM-ISAC, $48\%$ peak-picker ceiling). A peer-exclusion gate cuts peer reports from $4.4\%$ to $0.02\%$ of answered bursts, answering $23.5\%$ of total bursts ($75\%$ on the exact tap) with $0.70$ m/s velocity tracking. In-domain refitting recovers telemetry BER by up to three orders of magnitude under 3GPP channels, jamming-aware training restores the link, and frequency agility adds a $6$--$9$ dB margin against partial-band jamming.

cs.AI

新论文 · v1arXiv:2611.19525

Objectify, Don't Delete: Complaint-Derived Evaluation of Ambient Clinical Notes

Nikita Gupta, Narayanan Unny Edakunni, Vikram Okaly, Suranjana Datta, Navya Penubadi, Karl Eric Swanson, Siddharth Sharma

原文摘要

Evaluating the quality of AI-generated clinical notes is hard: standard metrics like ROUGE measure lexical similarity to reference notes, while expert checklists encode what committees believe matters. Neither captures what clinicians actually complain about. Recent work has shown that evaluation criteria can be mined from user feedback and filtered to criteria that LLM judges can reliably enforce. We argue that this is where much of the value is lost. The criteria that fail enforceability are disproportionately the ones clinicians raise most often, because high-frequency complaints are stated as several questions glued together. No yes/no judgment can grade such composites as posed. We introduce an objectification step that rewrites such criteria into sets of scoped, determinate decisions. We demonstrate the approach through an evaluation pipeline that mines criteria from complaints, objectifies them, and deploys LLM judges against both planted defects and production notes. Across eleven defects planted into ACI-bench reference notes, objectified judges localize violations with median recall of 0.95, while reference metrics change by at most 0.037. Given the complaint as one broad question, the same judges still find every planted defect, but they also flag 70% of the unperturbed reference notes; one section-pair question per call raises Youden's J from 0.30 to 0.72 (by 0.19 to 0.26 with prompt effort matched). We release ACI-Defect, a perturbation benchmark; clinician-reviewed instance-level labels showing quality gaps in the ACI-bench gold standard (32 of 40 notes contain hallucinations; 40 of 40 omit discussed facts); and grounding-judge verdicts. On 1,446 production notes, mean instance-rate burden falls at every clinician-rating step (rho=-0.15), a small association. Defect counts have to be kept per span rather than collapsed into one pass/fail bit per note.

cs.AI · cs.CL

版本更新 · v3arXiv:2601.08676

Advancing ESG Intelligence: An Expert-level Agent and Comprehensive Benchmark for Sustainable Finance

Yilei Zhao, Wentao Zhang, Lei Xiao, Yandan Zheng, Mengpu Liu, Wei Yang Bryan Lim

原文摘要

Environmental, social, and governance (ESG) criteria are essential for evaluating corporate sustainability and ethical performance. However, professional ESG analysis is hindered by data fragmentation across unstructured sources, and existing large language models (LLMs) often struggle with the complex, multi-step workflows required for rigorous auditing. To address these limitations, we introduce ESGAgent, a hierarchical multi-agent system empowered by a specialized toolset, including retrieval augmentation, web search and domain-specific functions, to generate in-depth ESG analysis. Complementing this agentic system, we present a comprehensive three-level benchmark derived from 310 corporate sustainability reports, designed to evaluate capabilities ranging from atomic common-sense questions to the generation of integrated, in-depth analysis. Empirical evaluations demonstrate that ESGAgent outperforms state-of-the-art closed-source LLMs with an average accuracy of 84.15% on atomic question-answering tasks, and excels in professional report generation by integrating rich charts and verifiable references. These findings confirm the diagnostic value of our benchmark, establishing it as a vital testbed for assessing general and advanced agentic capabilities in high-stakes vertical domains.

cs.AI

新论文 · v1arXiv:2611.19499

Do Probing and Steering Favor the Same Directions Within and Across Language Models?

Afshin Sadeghi, Daniel Becker

原文摘要

A direction in a language model's hidden state can be used to probe whether a concept is present or to steer the model toward it. Although probe directions are often reused for steering, the direction that best detects a concept need not be the one that best changes it. We study this distinction in three instruction-tuned models from different families. We construct a one-parameter family of directions ranging from a covariance-adjusted discriminant to the mean difference, and evaluate probing and additive steering, which adds the direction to the hidden state, along the family at a fixed side effect. For truthfulness and sentiment, in every model, the best probe is not the best steering direction: steering consistently favors directions closer to the mean difference. This separation holds across 59 of 60 layer-model-concept combinations in a layer sweep, with matched probe and steering texts in all six model-concept combinations, and under a rotation-based steering operator. The exact steering optimum is less stable: for sentiment, an intermediate direction produces a 27-56% stronger steering effect than the mean difference, but this gain shrinks or disappears with matched texts and depends on the layer and the operator. The two roles also transfer differently across models. A probe transferred by a linear map fitted on neutral text retains 81-91% of the target model's above-chance probing performance, while the transferred mean difference retains less of its steering effect in 11 of 12 model pairs. This is a consistent tendency, although our predefined criteria classify only three pairs, and nine remain inconclusive. Finally, as an application on PopQA, steering toward "unknown" increases abstention by 18-44 percentage points on uncertain errors, versus 2-5 on confident errors, while rarely correcting errors in cases where a correct answer is reachable.

cs.CL · cs.AI · cs.LG

新论文 · v1arXiv:2611.19472

InvarOrd: Self-Supervised Ordinal Knowledge Distillation with Rank-Invariant Alignment for Label-Efficient Cross-Species Hyperspectral Fish Freshness Assessment

Kazi Nabiul Alam, Omenuko Joseph Nwebonyi, Pooneh Bagheri Zadeh, Akbar Sheikh-Akbari

原文摘要

Hyperspectral models of fish freshness rarely transfer between species, and labelling a new species is slow and destructive. Freshness is also ordinal, and standard domain alignment can erase that order. We present InvarOrd, which adapts to a new species from 30 labelled cubes (5 per rank) in three stages: self-supervised spectral pretraining on unlabelled source cubes, cross-species alignment by a shrinkage covariance discrepancy computed only within matched ranks, and ordinal distillation into a compact student. The alignment term is provably invariant to the rank means, so it gives no incentive to collapse the ordinal axis. On paired salmon and cod data (800 cubes each, 16 days), a source-trained teacher reaches 1.6 to 1.9 days mean absolute error (MAE) on the other species, with quadratic weighted kappa (QWK) below zero. The 423K-parameter student, 2.54 times smaller than its teacher, attains the lowest MAE (1.476 days) and highest within 1-day accuracy (56.1%) among 14 configurations, matching the best of six ImageNet-pretrained backbones (ConvNeXt-T, 1.503 days) at 66 times fewer parameters, 1.44ms per cube on a GPU and 5s re-adaptation. InvarOrd beats target-only training by 0.29 days (p = 0.031). Among alignment objectives, rank-invariant alignment retains the most ordinal structure (embedding rank correlation 0.183, QWK 0.112), whereas global discrepancy lowers the rank correlation to 0.049 and QWK to zero at only 0.014 days higher MAE.

cs.CV · cs.AI

版本更新 · v3arXiv:2609.06563

MARBO: Relational Belief Grounding for LLM Agents in Social Deduction Games

Yechan Hwang, Sangjun Bae, Jeongmo Kim, Sangwoo Bang, Seungyul Han

原文摘要

Social deduction games (SDGs) require agents to reason under partial observability by maintaining relational beliefs about hidden roles and team alignments. While recent LLM-agent approaches improve gameplay through prompting and preference optimization, they often optimize actions and in-game speech without explicitly grounding them in such beliefs. This frequently leads to strategically inconsistent behavior, especially for compact LLM agents. We introduce Multi-Agent Relational Belief Optimization (MARBO), a belief-grounded preference optimization framework that leverages relational beliefs to guide strategic decisions and in-game speech. MARBO provides preference feedback only when behaviors are supported by reliable relational beliefs and lead to strategically favorable social outcomes, encouraging more consistent learning under uncertainty. Experiments on representative SDGs show that MARBO enables compact LLM agents to consistently outperform existing baselines. The Code is available on https://github.com/PleaseTakemeAway/MARBO.

cs.AI · cs.LG

新论文 · v1arXiv:2611.19465

VAST: Compiling Grounded Vulnerability Scenarios into Verifiable, Multi-Step Environments for Evaluating LLM Red-Team Agents

Zhen Wan

原文摘要

Evaluating Large Language Model (LLM) security agents requires reproducible environments whose workflow difficulty can be varied and whose failures can be diagnosed. We present VAST (Verifiable Adversarial Scenario Training & Evaluation), a declarative range compiler separating vulnerability scenarios from a compiler-introduced workflow-hardening layer. Following a post-hoc code and AST audit across all topics, the main benchmark retains 109 seed edges (45 with strict upstream CVE annotations), 72 family identifiers (62 under shared-HMAC relay grouping), 81 topic configurations (76 core plus 5 bounded execution models), and 1,387 instances (359 canonical four-gate-revalidated specs plus 1,028 seed variants) in a 42:39 partition. On the TopicServer compiler stratum, a budget-matched 270-episode prompt contrast across 10 compiled mechanism-family clusters (30 P2/P3/P4 instances, 40 turns, temperature 0.0) shows that glm-5.3, grok-4.7, and deepseek-v4-pro solve 12/90 instances under contract_only and 14/90 under generic_hint (cluster sign p=1.00), whereas supplying the shared workflow protocol in full_guide lifts completion to 76/90 (28/30, 23/30, 25/30; cluster sign p<=0.00391). On the separate ranges_v2 ledger, a 126-episode evaluation across 6 non-TopicServer workflows and multi-container Docker/Linux stacks solves 88/126 (69.8%) (28/42, 31/42, 29/42), while an 18-episode intervention on asynchronous Redis worker chaining (v2_02) lifts solves from 0/18 to 18/18 (100%) in 4-6 turns when installed worker modules are disclosed in GET /sched/directory. Finally, a three-seed (s in {42, 1024, 2025}) LoRA recipe comparison on Qwen2.5-Coder-7B-Instruct across 37 lineage-disjoint families (231 episodes on 33 held-out instances) yields 0/33 for Base, 1.67 +/- 2.89 / 33 for Golden SFT, and 4.67 +/- 2.31 / 33 for Header-Recovery SFT, with all solves on v2_03.

cs.CR · cs.AI

新论文 · v1arXiv:2611.19461

Program Distillation: Transferring LLM Capability from System 2 to System 0

Pujun Zheng, Wanying Ren, Guoxiu He

原文摘要

Recent studies leverage large language models (LLMs) to construct executable programs through environment interaction, feedback analysis, and code revision. However, prior work has mainly treated LLMs as optimizers, overlooking their ability to transfer knowledge across system levels. We propose Program Distillation (PD), a framework that formalizes the transfer of abilities from System 2 to System 0. During learning, the LLM provides task knowledge and draft strategies; at deployment, a student program executes the resulting policy. We show that this approach can improve performance while reducing inference time. We introduce PDBench, a benchmark comprising 27 discrete- and continuous-control tasks with fixed initial policies, and PDMetaHarness, a model-harness-independent meta-system for monitoring and guiding the learning process. The strongest model achieves a normalized mean score of 0.728, surpassing PPO (0.656) and Jev-like decision models (0.239), while running exclusively on CPUs and reducing inference costs. Our analysis shows that performance is largely driven by the first breakthrough during learning, and structural modifications have a slightly lower success rate than parameter modifications but typically yield larger gains. We further introduce style distillation, which enables interpretable stylistic control over the learning process. Overall, we present an analytical framework for PD and demonstrate its practical viability.

cs.AI

新论文 · v1arXiv:2611.19448

CLEW: Log-Anchored Backward Tracing for Root-Cause Code Localization

Wenji Bai, Muhammad Waseem, Jaakko Peltonen, Pekka Abrahamsson

原文摘要

Failures in industrial software systems may involve multiple components and be difficult to reproduce outside the production environment. Root-cause code localization therefore often relies on issue descriptions and runtime logs. However, log evidence may be noisy or incomplete, and the code reporting a symptom may differ from the code containing the underlying defect. These conditions make it challenging to identify relevant log evidence and determine which functions could have caused the reported failure. To address this gap, we introduce CLEW, an agentic approach that combines rule-based log-to-code anchoring with LLM-guided backward code tracing. CLEW organizes selected observations into diagnostic states, guides source exploration from anchored symptom functions, and returns ranked candidate functions, an affected module, and an explanation comprising a proposed mechanism path and associated log events. For evaluation, we construct LogLoc, a benchmark comprising 192 real-world issue and fault-injection cases. Against three baselines, CLEW records the highest observed Acc@1 (77.3%) and MRR (81.3%) on LogLoc-Issues and matches the leading baselines at 97.6% Acc@1 on LogLoc-Airflow. Averaged over all 192 cases, CLEW's wall-clock time is approximately 73% lower than LocAgent's at a similar average inference cost. No aggregate degradation is observed under the tested perturbations, with modest changes in Acc@1 and MRR across the evaluated seeds and base models. These results suggest that CLEW can combine function-level localization with evidence links and runtime efficiency under the evaluated conditions.

cs.SE · cs.AI

版本更新 · v2arXiv:2602.12083

Differentiable Modal Logic for Multi-Agent Diagnosis, Orchestration and Communication

Antonin Sulc

原文摘要

As multi-agent AI systems evolve from simple chatbots to autonomous swarms, debugging semantic failures requires reasoning about knowledge, belief, causality, and obligation, precisely what modal logic was designed to formalize. However, traditional modal logic requires manual specification of relationship structures that are unknown or dynamic in real systems. This tutorial demonstrates differentiable modal logic (DML), implemented via Modal Logic Neural Networks (MLNNs), enabling systems to learn trust networks, temporal orders, and regulatory boundaries from behavioral data. We present a unified neurosymbolic debugging framework through four modalities: epistemic (who to trust), temporal (what can follow what), deontic (what actions are permitted), and doxastic (who believes what). Each modality is demonstrated on concrete multi-agent scenarios (from exposing an attacker in a drone swarm to deciding when to consult another LLM code reviewer), with complete implementations showing how logical contradictions become learnable optimization objectives. Key contributions for the neurosymbolic community: (1) interpretable learned structures where trust and temporal order are explicit parameters, not opaque embeddings; (2) knowledge injection via differentiable axioms that guide learning with sparse data; (3) compositional multi-modal reasoning that combines epistemic and doxastic constraints; and (4) practical deployment patterns for monitoring, active control and communication of multi-agent systems. All code is provided as executable Python scripts.

cs.AI · cs.LO