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

+816新增论文
+642其中 AI 相关
+95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 13:49 — 2026-11-21 13:49(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:2611.23509

GRAFT: Execution-Guided Genetic Recombination for Agent Design

Sevda Ebrahimi, Mohammed Saidul Islam, Xiaodan Zhu, Elham Dolatabadi

原文摘要

Automated agent design searches for executable architectures that improve LLM-based agent performance, with evolutionary methods using language-model crossover to recombine promising mechanisms for task decomposition, model coordination, and verification. This search, however, is typically guided by aggregate fitness alone, which hides the instance-level successes and failures behind each score. We introduce Execution Guided Genetic Recombination for Agent Design (GRAFT), which uses aggregate fitness for selection and survival and parent-specific execution evidence to guide semantic crossover with integrated mutation. Its feedback harness reuses bounded samples of successful and failed runs from fitness evaluation, withholding reference answers and requiring no additional task executions. We evaluate GRAFT across six reasoning and programming settings using open-weight optimizer and executor models. It achieves a mean test accuracy of 82.0%, exceeding the strongest baseline by 9.8% points. Controlled ablations show that adding execution evidence improves test accuracy with initialization, search rules, repair policies, and evaluation budgets held fixed. Our results support the value of parent-specific execution evidence for evolutionary agent design.

cs.AI · cs.CL · cs.NE

新论文 · v1arXiv:2611.23508

BOS Summary: The First Token Matters

Hui Zhu, Saleh Zare Zade, Rafi Ibn Sultan, Deng Pan, Arun Ross, Dongxiao Zhu

原文摘要

Large vision-language models (LVLMs) and language models (LLMs) often exhibit attention sinks, whereby particular positions, such as the beginning-of-sequence (BOS) token, receive a disproportionate share of attention. Under standard causal masking, however, the BOS query attends only to itself, so the representation that later tokens repeatedly access carries no information about the input. We introduce BOS Summary, a training-free inference-time method that repurposes this persistent pathway to carry input context. At a single transformer layer during prefill, BOS Summary relaxes the causal constraint for the BOS query row alone, letting BOS aggregate all visual and textual input tokens into a self-derived context representation while every other token retains standard causal attention. The enriched representation then remains accessible to the model throughout decoding, with no additional parameters, auxiliary modules, or decoding-time overhead. Across multiple VLM and LLM families and eight multimodal and text-only benchmarks, BOS Summary improves over direct and Chain-of-Thought prompting and outperforms recent sink-intervention methods on multi-step reasoning tasks. A mechanistic analysis shows that making BOS input-specific also weakens its sink dominance and redistributes attention toward content-bearing prompt tokens, indicating that changing what a sink carries reshapes how later layers allocate attention.

cs.AI · cs.CL · cs.CV

版本更新 · v3arXiv:2512.15729

TinyMyo: a Tiny Foundation Model for Flexible EMG Signal Processing at the Edge

Matteo Fasulo, Giusy Spacone, Thorir Mar Ingolfsson, Yawei Li, Luca Benini, Andrea Cossettini

原文摘要

Objective: Surface electromyography (EMG) is widely used in biomechanics, rehabilitation, prosthetic control, and human-machine interfaces. Generalization across subjects, recording systems, and acquisition protocols remains challenging. Existing EMG foundation models (FMs) have limited support for multiple downstream tasks and deployment on low-power devices. Methods: We introduce TinyMyo, a 3.6-million-parameter Transformer encoder pre-trained on public EMG datasets using self-supervised masked reconstruction. TinyMyo supports multiple tasks with minimal changes to the task-specific prediction head. Results: We evaluated TinyMyo on hand gesture classification, hand kinematic regression, speech production, and speech recognition. For gesture classification, TinyMyo outperformed previously reported EMG FMs, achieving accuracies of 87.98% on NinaPro DB5, 97.1% on UCI-EMG, and 96.6% on EPN-612. On the GAP9 ultra-low-power microcontroller, TinyMyo achieved an inference time of 0.785 s and consumed 44.91 mJ per inference at an average power of 57.18 mW. A 1.9-million-parameter variant processed 5 s and 1 s input windows in 0.496 s and 0.089 s, respectively. Conclusion: TinyMyo demonstrates that a compact, self-supervised EMG encoder can support multiple downstream tasks and run on a low-power microcontroller. Significance: To our knowledge, this is the first deployment of an EMG FM on an ultra-low-power microcontroller, opening a path to on-device motor intent decoding, neuromuscular assessment, and human-machine interaction.

eess.SP · cs.AI · cs.HC · cs.LG

新论文 · v1arXiv:2611.23498

PhysDial: Adapting Pretrained Video Models for PDE Simulation under Changing Physical Conditions

Yilong Dai, Yiming Sun, Shengyu Chen, Jiaqi Gong, Xiaowei Jia, Runlong Yu

原文摘要

Simulating how physical systems respond to changing conditions is central to control, design, and scenario studies. To our knowledge, prior PDE foundation models have not been trained or evaluated on changes within trajectories. Our counterfactual test measures how far predictions move toward a changed solver branch from the same state. Models trained on fixed conditions follow changes only partly. A PDE foundation model given a condition input after pretraining barely responds. PhysDial adapts a pretrained video diffusion transformer to eight two-dimensional PDE families. Coefficients enter with the noise level, condition fields as reference frames, and equation text as a prompt. In Kolmogorov flow, switches within trajectories account for about 5% of the training samples. At matched compute, they halve the error after a change in forcing wavenumber or amplitude. A few trajectories at new amplitudes more than halve extrapolation error. Without physics pretraining, PhysDial matches or exceeds PDE foundation models under training conditions on two turbulent flows. At step 64 on the Kolmogorov test split, PhysDial has 1.07 times the solver's small-scale energy and Walrus 7.1 times. Our experiments suggest that a learned simulator uses a condition only as far as its training data make that condition necessary for prediction. Learning from trajectories that diverge under intervention can help scientific simulators predict what would happen if a system's forcing or parameters changed.

physics.flu-dyn · cs.AI

新论文 · v1arXiv:2611.23483

SpecLook: Speculating Where to Look for Fast and Reliable Agentic MLLMs

Xi Ding, Jiawei Zhang

原文摘要

Agentic multimodal large language models (MLLMs) that think with images interleave reasoning with zoom-in tool calls. However, this sequential loop is slow, and its answers carry no reliability score. Prior speculation of tokens, reasoning steps, or answers keeps the loop, and answer speculation also accepts confident errors. We find that the loop mostly adds one zoom-in step, with which a small model is ahead of the agent on most splits. Our key insight is to speculate the agent's next observation, which, unlike a drafted answer, can be verified, corrected, and used to test the answer. We propose SpecLook, a training-free speculate-and-verify framework with three levels. A small model drafts a glimpse and re-answers from it, and the question ends if the answer is unchanged and confident. Otherwise, the agent's large model verifies and corrects the glimpse without tools, and their answers are fused. Only questions that still lack evidence reach the agent, in one forward pass that SpecLook-Fast skips, so no question runs the loop. The trace scores every answer's reliability without labels. Extensive experiments on V*, HR-Bench, and POPE show that both are more accurate than every baseline on every split. SpecLook reaches 87.2% and 86.2% mean accuracy with 7.7x and 4.7x speedups over the DeepEyes-7B and Thyme-RL agents, and SpecLook-Fast reaches 9.9x and 6.3x.

cs.AI · cs.CL · cs.CV · cs.LG

新论文 · v1arXiv:2611.23480

Diagnostic Accuracy and Navigation Behavior of Vision-Language Model Agents Reading Emergency CT and MRI Examinations

Jonathan Lichtle, Aymen Meddeb, Antonin Birot, Ibrahima Niang, Lisa Gastaud, Riyad Hanafi, Sofiane Dujardin, Corentin Selomme 等 13 位作者

原文摘要

Purpose. To evaluate the diagnostic accuracy and image navigation behavior of vision-language model (VLM) agents under varying exploration strategies. Materials and Methods. In this retrospective study, six VLMs read 100 emergency CT and MRI examinations published in a public archive (March 2025-August 2026) under four conditions (free navigation, navigation from a key slice, and all slices or key slices in a single request), twice each (4800 runs). Nine radiologists read 30 of those cases. A large language model judge validated against two radiologists graded the answers. Conditions were compared with generalized estimating equations clustered by case. Results. Pooled accuracy was lowest under free navigation (32.6%; 391/1200), followed by navigation from a key slice (37.9%; 455/1200; P = .003), and highest with all slices (42.2%; 506/1200) or key slices (42.5%; 510/1200) in one request (both P < .001). On the 30 cases read by radiologists, freely navigating models were correct in 30.0% of runs (108/360) versus 77.0% for radiologists (208/270; P < .001). Models formed a first hypothesis after a median of 10.8% of the slices and kept it in 63.1% of runs (755/1196); final answers were correct in 12.4% of runs after a wrong first hypothesis (110/885) versus 90.4% after a right one (281/311; P < .001). Median stated confidence in the final answers was 90%, whereas 38.9% of them were correct (1861/4778). Conclusion. Across all exploration strategies, VLM agents remained far less accurate than radiologists. They adopted an early hypothesis and usually kept it, seldom correcting a wrong one.

cs.AI

新论文 · v1arXiv:2611.23476

GUARD: Stabilizing Recursive Self-Improvement with Regret Guidance

Minghe Zhang, Xuetao Kang, Yifei Ming, Yanfei Chen, Han Yu, Yufan Zhuang, Chen-Yu Lee

原文摘要

Autonomous AI agents pair a frozen base model with an editable harness of prompts, skills, tool rules, and memory. Recursive self-improvement (RSI) adapts this harness from experience without weight updates, yet self-editing without external graders often backfires: reflection and scalar scores conflate what an edit newly broke with what remains unsolved, causing local regressions, skill forgetting, and memory bloat. We introduce GUARD (Grader-free Unsupervised Environment Design [UED] with Adaptive Rule Distillation), a closed-loop framework for reference-free continual harness evolution. Our key insight is to decompose each task into pre-execution self-checks and guide harness updates via directional regret. Each round grades trajectories against these checks, pinning mastered rules while prioritizing newly broken behaviors, and selects among parallel candidate edits by critic-free group-relative fitness. A Soft-HURT gate then promotes an edit only when gains outweigh regressions, scoping specialized fixes to prevent over-triggering and guaranteeing non-regression on its gated buffer. At test time, one frozen harness runs in a single pass. Across benchmarks on memory, skill evolution, continual task streams, and domain reasoning, GUARD consistently improves performance without catastrophic forgetting or memory pollution while substantially cutting deployment token usage.

cs.AI · cs.LG

新论文 · v1arXiv:2611.23474

AllotNet: an Interpretable Mixture-of-Experts Choice Model for Moral Judgement in Humanitarian Aid Allocation, Validated against a Planted Synthetic Population

Mark Nowak

原文摘要

Official development assistance fell by 23.1 per cent in 2025, and the decision of who is helped first when there is not enough for everyone is moving into software. A leading published machine-learning targeting system excluded 53 per cent of eligible households as deployed. If the rule of distribution ends up in code, it should carry the judgement of many cultures, and the machine should execute while a person judges. AllotNet is an interpretable mixture-of-experts conditional logit for the allocation judgements elicited by Allot, a pre-registered six-language instrument: eight linear taste experts over seven ethical criteria, weighted by a gate over participant context, with a damped residual, an encoder for household intake, an outcome head and a budget-constrained planner that builds a review queue and never rejects anyone. A model of human judgement has no answer key on real data, so we wrote one first: a planted synthetic population of 5.6 million sessions and 104 million responses (98 million usable choices), generated by the production code path from a ground truth hidden from the loss. All sixteen pre-specified checks passed: criterion ranking $ρ= 1.000$, hidden segments recovered (ARI 0.671), outcome head at 99 per cent of the noise ceiling, planner at 99.4 per cent of oracle. The caveats are reported with them: two checks are weaker than their titles, and a live-generation failure cost 260,712 seeds. The model has 23,550 parameters, is trained quantisation-aware for NVFP4 with its interpretable core in fp32 (welfare-rank parity $ρ= 0.9937$ at W4A16 on the shipped bundle; one parity threshold was relaxed post hoc), and ships as a 125 KB bundle with C11, Java, Rust and Python runtimes. Nothing has been trained on real data yet; at the study's real scale a hierarchical Bayesian mixed logit is the registered comparator.

cs.LG · cs.AI · cs.CY

新论文 · v1arXiv:2611.23473

Reliability by Architecture: Agentic Workflows for Validated Multi-Physics Simulation from Natural Language

Søren Taverniers, Eric Darve, Hamdi A. Tchelepi

原文摘要

Preparing, executing, and validating strongly coupled multi-physics simulations require expert time. Agentic workflows use large language models (LLMs) to coordinate them from natural-language descriptions, but reliability requires checks of physical consistency and fidelity to the problem description. Our agentic harness implements explicit task allocation: LLMs interpret user requests and simulator diagnostics, generate input components not covered by predefined construction rules, and propose repairs, while deterministic routines construct supported components and check inputs, edits, and results. We evaluate CO2-storage problems using the multi-physics simulator GEOS, from compositional flow to flow-poromechanics with frictional fault contact. In 80 attempts on two flow problems, grounding raises the proportion of structurally valid inputs from 5/20 to 20/20, but acceptance reaches 20/20 only with complete rule-based input preparation. With identical inputs, local repairs correct three seeded defects while preserving valid inputs. Supporting tests identify simulations passing physics checks despite differing from the problem description or reference results.

cs.CE · cs.AI · physics.comp-ph

新论文 · v1arXiv:2611.24980

DL101 The Representation Capability of Neural Network Models

Fernando Berzal

原文摘要

Universal approximation theorems (Cybenko, Funahashi, Hornik & Leshno) provide the mathematical justification for why neural networks are powerful enough to model complex, nonlinear phenomena. Shallow networks can approximate any continuous function on compact sets. Leshno's theorem clarified that non-polynomial activations are necessary for universality, which explains why ReLU, logistic, hyperbolic tangent, and other common activations work (but pure polynomials do not). However, approximation theorems provide no efficiency, training, nor generalization guarantees. They do not indicate how many neurons are needed for a given accuracy, whether gradient descent will find the right parameters, or whether networks trained on finite data will generalize well. Barron provided quantitative error bounds for particular function classes (approximation rates for functions with bounded Fourier norm), whereas Montúfar was the first to show that expressivity grows exponentially with depth (in terms of the number of linear regions). Deep networks are strictly more expressive than shallow ones with the same number of parameters. Depth is fundamentally more powerful than width, providing a rigorous mathematical explanation for the use of deep learning models.

stat.ML · cs.AI · cs.LG · cs.NE

版本更新 · v3arXiv:2611.14804

GenPFN: An In-Context Tabular Foundation Model Approach for Augmenting Small Datasets in Engineering

Noah J. Bagazinski, Cyril Picard, Faez Ahmed

原文摘要

Costly simulations and experiments limit the data available for engineering design. This paper proposes GenPFN, a training-free candidate generator guided by a pretrained tabular Prior-Data Fitted Network (TabPFN) classifier with fixed weights. The context contains evaluated designs labeled desired (Class 1) or undesired (Class 0). GenPFN interpolates between desired designs and refines candidates using gated classifier gradients and Gaussian noise. Each candidate is evaluated using the problem's constraints and objective functions. Evaluated desired designs and their objective values augment the context for regression training. Evaluation spans fourteen selected problems, including a two-parameter benchmark and a 44-parameter hull problem. In paired comparisons, GenPFN with TabPFN v1 achieves higher class satisfaction and lower realism distance than the tested adversarial, variational-autoencoder and diffusion generators. The variational-autoencoder generator provides greater coverage and diversity. Relative to the synthetic minority over-sampling technique (SMOTE), GenPFN provides lower coverage distance and greater diversity. Among observed paired regression fits, augmentation reduces estimated AutoGluon validation mean squared error by 62.1% relative to context-only training. A hull-stern case study yields 99.3% propeller-compatible candidates versus 1.4% among randomly sampled context designs. Generation followed by evaluation produces sample sets that augment limited datasets and support training more accurate predictive models.

cs.AI

新论文 · v1arXiv:2611.23456

VeriMath: Human-in-the-Loop Generation and Verification of Curriculum-Aligned Mathematics Assessment Items

Quy P. Nguyen, Duong Q. Nguyen, Dinh P. T. Nguyen, Truong-Giang T. Le, Truong A. Bui, Vi H. Vo, Quang-Thinh Bui, My-Phuong Ngo 等 12 位作者

原文摘要

Automatic generation of mathematics assessment items is a growing area in educational artificial intelligence, yet classroom deployment remains challenging. Generated questions must satisfy multiple constraints: mathematical correctness, curriculum alignment, appropriate cognitive difficulty, and educator reviewability. While prior work has advanced text generation, reliable assessment authoring requires granular control over localized examination formats and expert educational judgment. To address this challenge, we present VeriMath, a generalized, human-in-the-loop framework for generating rigorous, curriculum-constrained mathematics assessment drafts. Using annotated reference items, the framework anchors generation to the requested examination section and cognitive reasoning level, then applies sequential automated verification and guided repair before submitting drafts for expert review. We evaluate VeriMath through a national curriculum case study using 4,555 mathematics items from the Vietnamese Grade 12 curriculum and 50 distinct generation requirements covering multiple-choice, true/false, and short-answer formats. Across two model backbones, the framework produces verifier-accepted drafts for 94% (47/50) of requirements per run. Blind expert evaluations confirm high mathematical accuracy and curriculum alignment, with strict teacher-readiness assessments accepting up to 56.8% of generated items. Analysis of rejected drafts reveals persistent domain-agnostic challenges involving formatting, statement validity, and solution-step completeness. These findings demonstrate that VeriMath can bridge generative AI and dependable educational applications by producing highly reviewable assessment drafts within tightly constrained national curriculum frameworks.

cs.AI · cs.CV

新论文 · v1arXiv:2611.23454

LusoClin: A Synthetic Clinical Note Dataset for Medical Information Retrieval in European Portuguese

Carla Teixeira Lopes, Daniel Félix

原文摘要

Access to large-scale clinical text collections is crucial for advancing medical information retrieval research, including clinical note search, patient cohort identification, and domain-adapted retrieval models. However, real clinical notes are rarely available due to privacy regulations, which particularly limits progress for low-resource languages such as European Portuguese. We present LusoClin, a publicly released dataset of synthetic clinical notes in European Portuguese, created to support privacy-preserving research on clinical text retrieval. LusoClin was generated by pre-processing real clinical data, fine-tuning the open-source Llama 3.2 (3B) instruction-tuned model using Quantized Low-Rank Adaptation, and applying prompt-based clinical note generation. To facilitate semantic indexing and clinical search, the notes are automatically annotated with clinical entities using a BERT-based model. We evaluate LusoClin in terms of utility, fidelity, diversity, and privacy using automatic metrics and human evaluation. Results indicate that the dataset captures clinically relevant language patterns and can support downstream IR applications such as clinical document retrieval and domain-specific model adaptation, although annotators can often distinguish synthetic from real notes and some medical inconsistencies persist.

cs.AI · cs.IR

版本更新 · v2arXiv:2606.04326

Measuring What Matters: Synthetic Benchmarks for Concept Bottleneck Models

Julian Skirzynski, Harry Cheon, Shreyas Kadekodi, Meredith Stewart, Berk Ustun

原文摘要

Concept bottleneck models predict using high-level concepts detected in inputs. Although these models provide a simple way to reap benefits from interpretability, they are hard to train and evaluate because very few real-world datasets include concept labels. In this paper, we develop synthetic benchmarks for concept bottleneck models. With our benchmarks, researchers can generate labeled datasets and control properties that affect performance, such as data modality, the choice of concepts, or annotation quality. They can thus find problems that suit these models and isolate the factors behind their failures. We demonstrate how our benchmarks can be used to evaluate major use cases for concept bottleneck models: decision support, i.e., helping human experts make better decisions; and automation, i.e., handling routine tasks without supervision. Our demonstrations show how the benchmarks can identify new failure modes and attribute them to either the architecture or the data.

cs.LG · cs.AI

新论文 · v1arXiv:2611.24979

Trace2ST: Recovering PLC Structured Text from I/O Traces with LLM Generation and Verification

Hsi Chang, Irfan Khan

原文摘要

Programmable logic controllers (PLCs), the industrial computers that operate power and process plants, often run programs whose source code is no longer available; however, every input and output of the controller is logged continuously. This paper therefore asks whether a large language model (LLM) can recover an executable program in Structured Text (ST), the standard PLC language of IEC 61131-3, from such input/output records (traces) alone and, equally important, how the recovered program can be verified. To answer this question, we build a generate-replay-feedback pipeline. First, the LLM writes a candidate program from a compressed summary of the trace, without ever seeing the original program. The candidate is then executed one control cycle (scan) at a time and compared with the recording, and whenever a mismatch occurs the corresponding part of the trace is returned to the LLM, for at most three revisions. Because every result depends on the software that executes the candidates, we first validate it against two independent open-source PLC runtimes and find no unexplained difference. On this basis, we evaluate 47 third-party programs from two public benchmarks. The recovered programs reproduce the recorded trace exactly in 43 cases and agree with the original on 94% of scans for unseen inputs. The remaining errors, however, lie mostly in program branches that the trace never exercised; once a trace that targets these branches is added, agreement on unseen adversarial inputs rises from 0.80 to 0.98. By contrast, a classical automata-learning baseline without an LLM fits the recording but falls to 0.40 on unseen inputs, which indicates that recovering program structure is the hard part. Finally, on real data from a public water-treatment testbed, the recovered control laws match unseen measurements closely ($R^2 > 0.99$) and expose the only actuator attack in the dataset.

cs.SE · cs.AI · cs.CY · cs.PL · eess.SY

版本更新 · v4arXiv:2607.11263

Bringing Back Rule Induction to Fluid Intelligence Research? An Initial Validation of the ARC-AGI Benchmark in Humans

Jasmin Thelen, Oliver Wilhelm

原文摘要

Two competing perspectives on fluid intelligence (gf) measures propose that performance is primarily constrained either by working memory capacity or by the ability to induce novel relations. The first perspective is currently dominant in measurement, as evident from the use of a limited set of recurring rules, whereas the second perspective is reflected in many definitions but rarely present in measurement. The ARC-AGI benchmark predominantly requires rule induction and was proposed as a measure of gf for both humans and artificial systems. However, its psychometric properties have not yet been examined in human samples. We therefore investigated the psychometric characteristics and nomological network of ARC-AGI in a first study with 100 participants. A compilation of ARC-AGI items showed good psychometric properties and correlated substantially with figural fluid intelligence as measured by a figural reasoning test ($ρ$ = .63). Associations with figural originality were weak. These findings provide initial support for the validity of ARC-AGI as a measure of human fluid intelligence. Future research should include more rule induction tasks as well as additional multivariate covariates. This study is unusual by studying a task in humans that was initially designed for machines. We suggest systematically embedding AI benchmarks into the nomological network of human cognitive abilities to enable more systematic evaluation and interdisciplinary cooperation.

cs.AI · cs.LG

版本更新 · v3arXiv:2509.10780

How K-12 Teachers Adapt Generative AI to Local Classroom Demands: Cultural Distance Between GenAI Defaults and Everyday Teaching Practices

Qing Xiao, Ruiwei Xiao, Hanqi Jane Li, Phenyo Phemelo Moletsane, Xinying Hou, John Stamper, Hong Shen

原文摘要

Generative AI (GenAI) is entering K-12 classrooms, but its default culture does not always fit local teaching. Existing accounts either treat this misfit as macro-level structural bias or examine it through single-case studies; neither captures how much work teachers must do to make outputs usable across tasks. This exploratory, theory-building qualitative study introduces cultural distance, the gap between GenAI's default cultural repertoire and the situated demands of teachers' practice. Through interviews with 30 K-12 teachers who frequently use GenAI, 10 each from South Africa, Taiwan, and the United States, we identified three levels defined by the effort teachers invested and whether it yielded a usable output: low (minor edits), mid (sustained prompting and revision), and high (structural gaps rooted in unsupported languages, undigitized knowledge, or policy filters that no teacher effort can close). We present cultural distance as a promising framework requiring further examination and draw implications for designers, policymakers, and educators.

cs.HC · cs.AI

版本更新 · v2arXiv:2611.13006

Measuring and Attributing Evidence Sensitivity in Medical Vision-Language Models via Symmetric Counterfactual Controls

Bhaskar Gurram

原文摘要

Medical vision-language models (VLMs) asked closed clinical questions ("Is there a polyp?") can score well by relying on the question rather than the image, and current hallucination benchmarks cannot separate true visual discrimination from editing shortcuts or shifts in the answer threshold. We introduce a Counterfactual Evidence Framework that pairs each lesion erasure with a shape- and tool-matched control edit elsewhere in the image, adds same-site ring controls, restored and partial erasures, degraded views and natural negatives, tests on held-out inpainting tools, separates discrimination (the Evidence Sensitivity Gap, ESG, and AUROC) from the answer threshold, and checks the labels with blinded physician readers and hidden-state probes. Across 507 main-test cases (1,115 views), nine frontier VLMs claim 41.1% to 61.6% of erased lesions (ESG 0.269 to 0.470), whereas MedScope-4B, a Qwen3.5-4B agent trained on automatically built counterfactual trajectories, claims 9.6% (ESG 0.758; p <= 0.001 in all 27 primary comparisons). Physician readers support automated erasure in colonoscopy, where frontier models still claim 45% of erased lesions against 1% for MedScope-4B, but call a finding present on 54.7% of pooled erased reads across all sources, which bounds what unverified erasure can certify. Across 12 pre-registered training or system arms and 270 decision policies screened under a rule fixed in advance, every apparent utility gain comes from shifting the threshold rather than improving held-out discrimination, and probes point to perception as the remaining limit. We also report a 0/48 pneumothorax failure on an external cohort; retraining with pneumothorax patients in every stage finds 53% of pneumothoraxes on new patients without changing the main-test results.

cs.AI · cs.CL · cs.CV

版本更新 · v2arXiv:2609.26105

Neoadjuvant chemotherapy response prediction using pretreatment diffusion and contrast-enhanced magnetic resonance imaging with clinical variables

Pablo García Marcos, Paula Puerta González, Guillermo Lorenzo, Héctor Gómez, Covadonga del Camino, Adán Rodríguez, Ignacio Peláez, Angel Rio-Alvarez 等 9 位作者

原文摘要

Prediction of pathological complete response before neoadjuvant chemotherapy may facilitate more tailored therapeutic planning for breast cancer patients. This work proposes a deep-learning model for pretreatment data only, combining apparent diffusion coefficient maps, dynamic contrast-enhanced magnetic resonance imaging, and clinical variables. The study uses the public ACRIN 6698/I-SPY2 multicenter dataset. The architecture employs EfficientNet-B0 pretrained encoders for image feature extraction and late fusion with clinical information. Multiple clinical variables were evaluated, including age, race, histological type, HR/HER2 subtype, SBR grade, and maximum diameter. Only HR/HER2 subtype improved the average area under the receiver operating characteristic curve (AUC) and was retained in the final model. Using stratified five-fold cross-validation, standalone apparent diffusion coefficient maps achieved a mean AUC of 0.79, whereas dynamic contrast-enhanced magnetic resonance imaging achieved 0.74. Adding HR/HER2 subtype improved performance to 0.83 and 0.81, respectively. The final configuration, using both imaging modalities and HR/HER2 subtype, achieved an AUC of 0.86. These results support pretreatment multimodal learning for response prediction, although external validation is required before clinical use.

cs.AI · cs.CV

版本更新 · v2arXiv:2606.12563

Arbor: How Search Policy Shapes Agent Behavior in System Optimization

Neha Prakriya, Chaojun Hou, Zheng Gong, Huasha Zhao, Yanyuan Qin, Chris Cai, Xi Zhao, Zhenyu Gu 等 9 位作者

原文摘要

Achieving peak performance on modern accelerators requires coordinated optimization across kernels, operator fusion, KV-cache management, quantization, scheduling, and communication overlap, a process that demands weeks of expert effort per model and hardware platform. We present Arbor, a system that gives autonomous agents an explicit optimization tree to reason over, backtrack through, and deepen with experience, with pluggable search policies, scored bandit traversal, parallel agent swarms, and depth-first sweep, sharing a common harness of containerized isolation, automatic rollback, and end-to-end validation. Arbor generalizes across domains: +40% to +193% on six production inference models across two GPU generations, +42.8% on LLM training, and +41% on recommendation training, where agents autonomously wrote kernel-level and framework-level optimizations spanning thousands of lines of code. In controlled ablations where every search policy receives the same agent, tree, and tools, we find that search structure shapes both the magnitude and the nature of the optimizations. The bandit achieves +30.1% through targeted framework fixes, the sweep +26.4% through deep kernel rewrites, and the swarm +33.4% by compounding independent changes. An agent-driven loop that receives the same harness, tools, and information, but selects its own targets reaches +14%, writing 4x more code per percentage point of improvement than Arbor's structured search policies. We additionally introduce an evaluation framework for autonomous loop-based agents spanning efficiency, grounding, robustness, and depth, designed to characterize agent search behavior across long-running optimization campaigns.

cs.AI

版本更新 · v2arXiv:2611.20011

ReactBench-AD: Do Driving VLAs and World-Action Models React to Imminent Hazards in Their Path?

Nikos Theodoridis, Reenu Mohandas, Sushil Sharma, Ciarán Hogan, Ganesh Sistu, Anthony Scanlan, Tim Brophy, Ciarán Eising

原文摘要

Vision-Language-Action models (VLAs) and World-Action Models (WAMs) have emerged in autonomous driving with the promise of strong generalization to long-tail scenarios. Although these models achieve state-of-the-art results on autonomous driving benchmarks, most existing benchmarks are dominated by nominal scenarios and therefore cannot test this promise. To test one such long-tail capability that is critical for safety, we introduce ReactBench-AD, a benchmark that measures how driving models react to objects that appear in the ego vehicle's path. Built on real NAVSIM and PhysicalAI-Autonomous-Vehicles scenes, ReactBench-AD composites a crossing pedestrian, a standing pedestrian, or a stationary car consistently across all camera views and history frames. Each edited scene thus forms an exact, pixel-level counterfactual pair with its original, which allows us to isolate the causal effect of the inserted object on the predicted trajectory. Evaluating four VLAs and two WAMs, we find that they react weakly, often barely at all. Finally, we test how easily one of the evaluated VLAs, ReCogDrive, can be taught to brake hard for imminent hazards: a short fine-tuning stage on edited scenes substantially improves its reactions, with the learned behavior generalizing to unseen objects and fully real scenes, at minimal cost to nominal driving. These results call into question the out-of-the-box long-tail capabilities of these models, but suggest that they can be a strong foundation for acquiring them with little additional training, highlighting the value of real or synthetic data that demonstrates the long-tail behaviors nominal datasets rarely contain.

cs.CV · cs.AI · cs.RO

版本更新 · v2arXiv:2610.33122

The Digital-Energy Recomposition of the Kitchin Cycle: Exploratory Evidence from U.S. IT and Irish Data-Centre Series (1995-2025)

Bartosz Ziółko, Adam Niedzialkowski, Bartosz Szyma

原文摘要

Using quarterly FRED and CSO Ireland data over 1995Q1-2025Q4, we examine whether U.S.\ IT investment and Irish data-centre electricity are recomposing the Kitchin cycle around energy- and compute-backed assets. Three findings stand out. First, a Quandt-Andrews sup-Wald test locates a significant break in the U.S.\ IT-GDP correlation at 2014Q3 (sup-W $= 9.69$; $r$ falls from $+0.45$ to $-0.12$); the Fisher $r$-to-$z$ test survives Bonferroni adjustment ($z = +3.11$, $p_{\text{Bonf}} = 0.006$), whereas the analogous test on the imposed Cloud-vs-AI boundary does not ($p_{\text{Bonf}} = 0.164$), so ``Strategic Decoupling'' pre-dates the AI Era by approximately six years. Second, in Ireland the synchronous AI-Era correlation between metered data-centre electricity and Real GDP is large ($r = +0.41$, $N = 16$) but \emph{collapses to $r \approx 0$ when Real GDP is replaced by Real GNP}, suggesting the headline is largely an artefact of the shared multinational tax-structure footprint between hyperscaler operations and Irish gross output. Third, the Irish cross-correlation profile is asymmetric in the opposite direction to a leading-indicator reading: the largest positive coefficients cluster on the $k < 0$ side (GDP precedes data-centre load), so compute capacity is better read as coincident or moderately lagging. Given the small AI-Era samples and the absence of a quarterly multinational-adjusted aggregate such as GNI*, all coefficients are exploratory; subject to these caveats, the U.S.\ Kitchin cycle is best dated to 2014Q3, and the Irish Energy-Data Nexus is dominated by multinational firm activity.

physics.soc-ph · cs.AI · econ.GN · q-fin.CP · q-fin.GN

版本更新 · v4arXiv:2610.17895

Interaction-as-a-Judge

Wayne Chi, I-Hung Hsu, Li Chen, Wei-Lin Chiang, Ion Stoica, Anastasios Nikolas Angelopoulos, Ameet Talwalkar, Chris Donahue 等 9 位作者

原文摘要

Coding agents are tasked with producing sophisticated interactive applications such as websites, desktop applications, and games. While humans assess these applications interactively, current automated LLM evaluations primarily focus on code analysis and unit tests. This mismatch can cause LLM judges to miss failures that are immediately apparent to human users. We study Interaction-as-a-Judge, an evaluation paradigm where computer-use agents (CUAs) directly interact with an application. We demonstrate that Interaction-as-a-Judge can improve both evaluation and generation of interactive applications. To this end, we collect and release two complementary datasets: $6,858$ pairs of applications with human preferences and corresponding agent traces, and $222$ applications with human-verified requirement labels. Relative to standard non-interactive approaches, we find that Interaction-as-a-Judge improves evaluation in terms of predicting human preferences ($65.6\% \to 69.1\%$) and requirement verification ($77.0\% \to 81.6\%$). We also observe scaling benefits: across both preference alignment and requirement verification, higher-quality interactions improve performance. Finally, through both inference- and training-time experiments, we demonstrate that Interaction-as-a-Judge can also improve the quality of generated artifacts. At inference time, Interaction-as-a-Judge repairs $75.5\%$ of broken requirements within five attempts, compared to $55.4\%$ for code-only self-feedback. At training time, using Interaction-as-a-Judge as a reward signal improves win-rate over the base model by $6.7$pp while reducing both generated code length by $35.7\%$ and build failure rates by $7.2$pp compared to a code-only reward signal.

cs.LG · cs.AI · cs.SE

版本更新 · v3arXiv:2605.19049

KVBuffer: IO-aware Serving for Linear Attention

Longwei Zou, Lin Zhong

原文摘要

Linear attention has recently gained significant attention for long-context inference due to its constant decoding cost with respect to context length. However, existing serving systems typically serve linear attention by recurrently computing and updating a large linear attention state in every decoding step. Since the state is much larger than the per-token key and value, recurrent decoding incurs substantial memory access and becomes inefficient for serving linear attention. In this paper, we propose KVBuffer, an IO-aware serving mechanism for linear attention. By buffering recent keys and values, KVBuffer enables serving systems to compute linear attention outputs in more flexible and memory-efficient ways. For decoding, KVBuffer enables chunkwise computation, which reduces average memory access and decoding latency by deferring state updates and applying them in batch. For speculative decoding, KVBuffer verifies draft tokens in parallel and avoids storing temporary states. For short contexts, KVBuffer computes attention outputs directly from buffered keys and values, without creating or updating the linear attention state. We implement KVBuffer in SGLang for Qwen3.5 and Nemotron-3 series models. Our evaluations show that KVBuffer can speedup batch decoding by up to 1.47x and increase the maximum number of serving requests by 5x for speculative decoding when verifying four draft tokens.

cs.LG · cs.AI

新论文 · v1arXiv:2611.23410

RFBench: A Simulator-Grounded Multimodal Benchmark for Passive RF Layout Design and Understanding

Jiarui Cui, Boxun Yan

原文摘要

While multimodal language models can interpret diagrams and generate structured outputs, their capacity to connect passive radio-frequency (RF) design intent, layout geometry, and electromagnetic behavior remains largely untested. To address this gap, we introduce RFBench, a simulator-grounded dataset and benchmark for passive RF layout design and understanding. The dataset contains 10,000 two-port planar layouts simulated using ADS Momentum, with each layout represented by editable geometric primitives, a raster image, a 16x16 occupancy grid, complex S-parameters at 201 frequency points, a structural category, and an RF response label. Building on these aligned representations, we define seven tasks spanning specification-conditioned layout generation, representation translation, response prediction, simulator-based validation, design diagnosis and improvement, structural understanding, and response classification. Extensive baselines across seven multimodal language models reveal systematic weaknesses in generating structurally valid layouts and predicting electromagnetically faithful responses. We show that RF-specific fine-tuning increases T1a structural validity from 0.0% to 83.2% and the T3 primitive-input within-2-dB rate from 17.7% to 62.8%, demonstrating the value of domain adaptation while leaving considerable room for further progress. RFBench establishes a common testbed for evaluating multimodal RF capabilities and guiding the development of models that connect design intent and layout representations with physical behavior.

cs.AI · cs.CV · cs.LG

新论文 · v1arXiv:2611.23403

Tracking Representations in LLM Reasoning Traces for Iterated Games

Chaipat Tirapongprasert, Rajarshi Das, Parijat Dube

原文摘要

Large language model (LLM) agents achieve state-of-the-art performance across domains through chain-of-thought (CoT) reasoning. Despite their ubiquitous use, little is understood about the relationship between the externalized trace and the model's actual computation. We address this by investigating how LLM agents make decisions in iterative games with incomplete information. Using a series of mechanistic interpretability techniques, we find that the model forms and iteratively revises an internal representation of the game throughout the CoT. Once the model prematurely commits to an incorrect belief, it rarely recovers. This failure is manifested in both the literal trace and the layer activations. Ultimately, as multi-agent systems are more prevalent, our work opens the door to a more focused diagnosis of their strategic reasoning errors. For illustrative purposes, we offer an interactive visualizer of our methodology on selected sample traces at https://interp-self.github.io/gametrace/

cs.CL · cs.AI · cs.GT

新论文 · v1arXiv:2611.23398

HardwareDx: A Foundational Graph Model System for Datacenter Machine Repairs

Yuanlin Wen, Elan Markowitz, Zubo Gu, Sami Abu-El-Haija, Bryan Perozzi, Mikhail Galkin, Ali Parviz, Aayush Gupta 等 14 位作者

原文摘要

The rapid scale-up of AI infrastructure has made datacenter reliability a cornerstone of the cloud business economy. Since hardware failures in hyper-scale AI clusters can stall multi-million dollar training workloads, the downtime of individual machines can cause outsized economic losses. Minimizing machine downtime, therefore, is more important than ever. To address this, we present Hardware Diagnoser (HardwareDx), an AI-driven hardware diagnosis system that applies a graph foundation model approach to automated datacenter maintenance. It allows a single model to diagnose many platforms and accelerate onboarding of new hardware products. Deployed at production scale at Google, HardwareDx yields substantial efficiency gains across diverse compute architectures. Real-world evaluations demonstrate that it reduces production downtime by 18.3% on GPU platforms and 16.1% on TPUs. Furthermore, it significantly accelerates hardware deployment, cutting quality assurance (QA) time by 28.3% for GPUs and 25.1% for TPUs. These results establish HardwareDx as a field-tested solution for maximizing the reliability and cost efficiency of next generation AI fleets.

cs.AI · cs.AR

新论文 · v1arXiv:2611.23397

Which Tokens, in What Order, and How Fast? A Theoretical Analysis of On-Policy Distillation Objectives

Hubert Strauss, Liam Fowl, Noam Razin

原文摘要

On-policy distillation (OPD) trains a student language model to match a teacher model based on its own generations. When using OPD, one must choose whether to apply it at the sequence or token level and which divergence to minimize. However, despite the wide adoption of OPD in practice, the formal understanding of which OPD objective is appropriate for a given setting is limited. Toward addressing this gap, we theoretically analyze how the choice of OPD objective governs which tokens are prioritized, in what order, and how efficiently. Our analysis reveals that sequence-level OPD exhibits a future agreement bias, which myopic token-level objectives avoid. This bias induces a reverse learning order, in which sequence-level OPD first matches the teacher at later token positions, while at earlier positions it can favor tokens that are improbable under the teacher, and thus often undesirable. As for the choice of divergence, we prove that it strongly influences the amenability of OPD to gradient-based optimization. In particular, while the forward KL divergence enjoys a well-behaved objective landscape, other $f$-divergences, such as reverse KL, can suffer from vanishing gradients even when the student and teacher are far apart. Across language model post-training settings, we demonstrate the implications of these theoretical findings for which OPD objective is preferable, highlighting the importance of accounting for the interplay between the student, teacher, and task at hand.

cs.LG · cs.AI · stat.ML

新论文 · v1arXiv:2611.23388

QDA: Closed-Loop LLM Quantization Design with Agents

Xingyang Li, Tianlang Zhao, Song Han, Ligeng Zhu

原文摘要

Post-training quantization is the standard way to serve large language models at low cost. However, the best quantization recipe differs across models and precisions. As a result, experts must redesign it for every new model, which is time-consuming and often sub-optimal. Fortunately, an LLM agent can take over this design by refining the recipe over rounds of building and evaluating candidates, so that more rounds lead to better recipes. However, evaluating one quantized reasoning model takes tens of GPU-hours, and combining methods from different codebases means porting each to the new architecture by hand. We present QDA, a system in which an agent designs the recipe for each model and precision. To make evaluation fast, QDA scores each candidate in minutes by its output KL divergence from the original model on reasoning traces that the agent never sees, and checks promising ones with long generations to catch failures such as repetition loops. To make building fast, QDA provides a library of public PTQ methods that are compatible with one another and with new model architectures. Across four reasoning models, QDA improves accuracy over the strongest public recipe by 0.9, 1.4 and 2.2 percentage points on average at W4A16, W4A8 and W3A16, using 64 GPU-hours per search. On the 320B-parameter MoE model GLM-5.3-Flash at NVFP4 W4A4, it recovers 98.3% of the FP8 model's accuracy, against 96.9% for the best vendor checkpoint. Every QDA checkpoint also runs on existing serving kernels with no inference overhead. We will release the library, the evaluator and the harness.

cs.LG · cs.AI · cs.CL

新论文 · v1arXiv:2611.23375

Where the Tokens Go: Heat, Entropy and Error in Agentic Coding

Alán Aspuru-Guzik, Varinia Bernales

原文摘要

Coding agents are billed by the token, yet software engineering says nothing about what a token actually buys. Here we introduce a theory of software thermodynamics. The theory draws analogues from classical and stochastic thermodynamics, extends them, and applies them to modern agentic coding. We count the tokens an agent generates as its cost and divide that cost into work and heat. The test suite acts as a software Maxwell's daemon that judges every coding change. We test the theory with an analogue of the experiments in which a single molecule is pulled open and let fold again: an agent makes a change to a repository (coding) and then undoes it (uncoding), many times over, and we count the tokens of every model call. The cycle returns the requested edits to where they began, with each direction started from a stored reference tree, and the spread of its cost over repeated cycles defines a software temperature, which rose with the sampling temperature of the model. About 80 per cent of the tokens of a successful run are generated in model calls that leave the requested change where it was, and we call these tokens heat. The free energy of a change, which we estimate by analogy with the equality of Jarzynski, is 14 to 22 per cent of the mean cost of a successful coding run. Coding costs 1.1 to 1.5 times as much as uncoding. The daemon is imperfect: the verdict of a test suite varies with its environment, and a suite that accepts a broken change has a cost in tokens. The practical implication of the proposed theory could be new tools for analysing and optimising coding agents through the lens of thermodynamics.

cs.AI · cond-mat.stat-mech · cs.SE