跳到正文
研踪PaperTrail

版本更新

查看已发布论文的最新修订,在详情页对照实际观测到的版本。

部分采集异常

全站更新概况

+1084新增论文
+883其中 AI 相关
+130已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-18 04:28 · 全站最近入库:2026-11-18 19:45各信源进度 →

统计口径与数据新鲜度

统计 2026-11-18 05:21 — 2026-11-19 05:21(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:2026-11-18 04:28最近发现新增版本:2026-11-18 19:45库内论文最新更新时间:2026-11-18 08:00

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。

26,870 已收录论文22,374 AI 相关8,156 有修订112 HF 社区发现

论文修订列表

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 8,156 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-18 04:27(北京时间)
版本更新 · v2arXiv:2611.01301

CuriousTogether: Implications of Curiosity and Cultural Grounding in Multi-Agent LLM Interactions

Angana Borah, Zoha Hayat Bhatti, Zhijing Jin, Rada Mihalcea

原文摘要

Large Language Models (LLMs) are increasingly deployed in multi-agent settings, exhibiting emergent collective behaviors, yet the role of curiosity as a driver of such dynamics remains understudied. We introduce CuriousTogether, a framework evaluating curiosity-driven multi-agent LLM interactions across three complementary dimensions: novelty-seeking, epistemic and social curiosity. Across eight tasks spanning reasoning, creative generation, and social bias mitigation, curiosity-guided multi-agent settings yield improvements over non-curious ones on seven out of eight tasks, with gains of up to 20.6% in story generation (semantic diversity) and a 14.9 pp reduction in stereotype rates. Furthermore, country-conditioned cultural grounding show benefits for story generation and social tasks, but can hurt precision-oriented reasoning. Finally, we show that inference-time multi-agent curiosity-based approach offers favorable quality-token trade-offs, compared with single-agent baselines. Together, these findings support curiosity-guided interaction as a practical mechanism for improving multi-agent collaboration, with downstream benefits for reasoning, creative, and societal tasks.

cs.CL · cs.AI

版本更新 · v3arXiv:2602.24279

A quadratic lower bound for 2DFAs against one-way liveness

Kehinde Adeogun, Christos Kapoutsis

原文摘要

We show that every two-way deterministic finite automaton (2DFA) that solves one-way liveness on height h has Omega(h^2) states. This implies a quadratic lower bound for converting one-way nondeterministic finite automata to 2DFAs, which asymptotically matches Chrobak's well-known lower bound for this conversion on unary languages. In contrast to Chrobak's simple proof, which relies on a 2DFA's inability to differentiate between any two sufficiently distant locations in a unary input, our argument can be applied to inputs over any alphabet and is structured around a main lemma that is general enough to potentially be reused elsewhere.

cs.FL

版本更新 · v3arXiv:2605.31281

Wind Turbine Maintenance Log Labelling Framework: LLM-Driven Data Correction and Enrichment via Semantic Extraction of Reliability Intelligence

Max Malyi, Jonathan Shek, Alasdair McDonald, Andre Biscaya

原文摘要

Inconsistent codes and unstructured text limit the use of wind turbine maintenance records in reliability analysis. A topology-aware large language model (LLM) workflow corrected and enriched 16,316 records from 280 turbines at 31 onshore wind farms over 9.2 years. Maintenance narratives and equipment context guided system-code correction and maintenance-type relabelling. System- and topology-specific batches yielded candidate action and failure-mode taxonomies for record-level labelling. Failure-mode profiles linked labels to common mechanisms, possible symptoms and candidate causes. Schema constraints, deterministic rules and source links supported auditable processing. Proposed labels required model-reported High confidence and no human-review flag. Accepted system-code updates covered 2,178 of 2,984 targeted records; accepted failure-mode labels covered 11,661 records (71.5%). The enriched fields separated pitch interventions from broader system categories and distinguished electrical and hydraulic pitch maintenance patterns. Recorded task durations totalled 6.83 hours. A source-based assessment covered 800 records selected by balanced probability sampling. Changed system codes and maintenance types required source support; retained specific labels were approved unless contradicted. Equal-weight mean approval across four fields was 86.3%: 94.8% for system codes, 87.3% for maintenance types, 85.6% for actions and 77.5% for failure-mode outcomes. Failure-mode assessment covered titles and unresolved, unmatched or inapplicable statuses. Qualitative reviews identified unsupported actions and unclear category boundaries. The structured records provide traceable evidence for engineering review and future operational linkage supporting event reconstruction, exposure-based reliability analysis and failure modes and effects analysis.

cs.CL

版本更新 · v3arXiv:2505.24030

From Images to Signals: Are Large Vision Models Useful for Time Series Analysis?

Ziming Zhao, ChengAo Shen, Hanghang Tong, Dongjin Song, Zhigang Deng, Qingsong Wen, Jingchao Ni

原文摘要

Transformer-based models have gained increasing attention in time series research, driving interest in Large Language Models (LLMs) and foundation models for time series analysis. As the field moves toward multi-modality, Large Vision Models (LVMs) are emerging as a promising direction. In the past, the effectiveness of Transformer and LLMs in time series has been debated. When it comes to LVMs, a similar question arises: are LVMs truely useful for time series analysis? To address it, we design and conduct the first principled study involving 4 LVMs, 8 imaging methods, 18 datasets and 26 baselines across both high-level (classification) and low-level (forecasting) tasks, with extensive ablation analysis. Our findings indicate LVMs are indeed useful for time series classification but face challenges in forecasting. Although effective, the contemporary best LVM forecasters are limited to specific types of LVMs and imaging methods, exhibit a bias toward forecasting periods, and have limited ability to utilize long look-back windows. We hope our findings could serve as a cornerstone for future research on LVM- and multimodal-based solutions to different time series tasks.

cs.LG · cs.AI · cs.CV

版本更新 · v2arXiv:2605.17778

Self-Distillation is Optimal Among Spectral Shrinkage Estimators

Radu Lecoiu, Debarghya Mukherjee, Pragya Sur

原文摘要

Self-distillation has emerged as a promising technique for improving model performance in modern machine learning systems. We develop the statistical foundations of self-distillation in high-dimensional linear regression by introducing and analyzing a broad class of estimators, namely spectral shrinkage estimators. We consider both spiked and general covariance matrices. For the former, we establish that when the covariance has $s$ distinct spikes, $s$-step self-distillation achieves optimal prediction among spectral shrinkage estimators, outperforming well-known estimators in statistics and machine learning. Moreover, we show that $s$ steps are necessary for optimality: any $(s-k)$-step distilled estimator is strictly suboptimal for $1 \leq k \leq s$. For general covariance matrices, we show that self-distillation approaches the optimal limiting prediction risk among continuous spectral shrinkage estimators as the number of distillation steps increases. We further develop a generalized cross-validation procedure for selecting the distillation parameters from data and establish an asymptotic oracle guarantee for the resulting estimator. We also study a federated approach where multiple data centers share spectral shrinkage estimators and a common server seeks to aggregate them to achieve optimal performance. In this case, we find that the best local rule again takes the form of self-distillation, though it differs from the optimal rule when data are hosted centrally on a single server. Together, our results explain why self-distillation improves predictive performance and provide a broader statistical framework that connects it to classical shrinkage-based methods.

math.ST · cs.LG · stat.ME · stat.ML

版本更新 · v2arXiv:2602.10039

Budgeting Discretion: Theory and Evidence on Street-Level Decision-Making

Gaurab Pokharel, Sanmay Das, Patrick J. Fowler

原文摘要

Street-level bureaucrats, such as caseworkers, triage nurses, and school administrators, operate under a fundamental tension. Official policy on who to prioritize may conflict with bureaucrats' estimation of who would benefit most. The theory of street-level bureaucracy argues that frontline agents bridge this gap, overriding policy when professional judgment identifies opportunities to improve outcomes. As algorithmic recommendations enter allocation settings, understanding this behavior is key to designing tools that account for professional judgment. We formalize discretion as a finite-horizon dynamic allocation problem. An agent follows a default policy but has a budget of $K$ overrides across $T$ periods. Each period, the agent observes the welfare gain from deviating and decides whether to spend an override now or save it for later. We show that the optimal policy is a state-dependent threshold rule and prove a behavioral invariance: for location-scale families of improvement distributions, the rate at which an optimal agent exercises discretion is independent of welfare-gain magnitude and depends only on the distribution's shape. Fat-tailed gains induce patience (conserving overrides for rare, high-stakes cases); thin-tailed gains induce routine, front-loaded spending. We also compare the model's implications with data from a homelessness service system. In the model, a capacity opening makes an override cheaper to spend, lowering the threshold for using one. In the data, override rates shift with short-run capacity openings in scarce housing, rising when exits free slots and falling when interventions are congested. Descriptively, spending is front-loaded: override rates peak on Mondays and early in the fiscal year. Together, the model and data point to systematic structure in discretion, informing the design of more efficient and equitable human-AI allocation systems.

cs.CY

版本更新 · v2arXiv:2603.26097

Dynamic Tokenization via Reinforcement Patching: End-to-end Training and Zero-shot Transfer

Yulun Wu, Sravan Kumar Ankireddy, Samuel Sharpe, Nikita Seleznev, Dehao Yuan, Hyeji Kim, Nam H. Nguyen

原文摘要

Efficiently aggregating spatial or temporal horizons to acquire compact representations has become a unifying principle in modern deep learning models, yet learning data-adaptive representations for long-horizon sequence data, especially continuous sequences like time series, remains an open challenge. While fixed-size patching has improved scalability and performance, discovering variable-sized, data-driven patches end-to-end often forces models to rely on soft discretization, specific backbones, or heuristic rules. In this work, we propose Reinforcement Patching (ReinPatch), a novel framework that jointly optimizes a sequence patching policy and its downstream sequence backbone model with reinforcement learning. By formulating patch boundary placement as a discrete decision process optimized via Group Relative Policy Gradient (GRPG), ReinPatch bypasses the need for continuous relaxations and performs dynamic patching policy optimization in a natural manner. Moreover, our method allows strict enforcement of a desired compression rate, freeing the downstream backbone to scale efficiently, and natively supports multi-level hierarchical modeling. We evaluate ReinPatch on time-series forecasting datasets, where it demonstrates compelling performance compared to state-of-the-art data-driven patching strategies. Furthermore, our detached design allows the patching module to be extracted as a standalone foundation patcher, providing the community with visual and empirical insights into the segmentation behaviors preferred by a purely performance-driven neural patching strategy.

cs.LG · cs.AI · stat.ML

版本更新 · v2arXiv:2606.04130

CLAW: Continuous Latent Action World Models via Adversarial Regularization

Tewodros Ayalew, Samuel Wheeler, Matthew Jeung, Xiao Zhang, Andre de la Cruz Arce, Kaylene Stocking, Michael Maire, Matthew R. Walter

原文摘要

We introduce CLAW, a fully end-to-end self-supervised framework for learning a world model jointly with continuous latent action representations directly from action-free videos. Our approach leverages adversarial latent regularization to prevent the latent from encoding future observations, so that it captures the structure of the action. The diffusion-based world model then models environment dynamics as a distribution over future observations rather than as a single deterministic rollout. Both components are learned from observation sequences alone without any action annotations. By simultaneously training the Latent Action Model and world model, CLAW learns to reason about how inferred actions induce environment transitions from visual observations alone. We show that the resulting latent action world model supports both imitation learning from observation and goal-directed planning. In imitation learning, latent actions extracted from raw videos enable behavior cloning. For planning, CLAW generates sequences of latent actions and maps them to executable actions to reach desired goals. Extensive experiments across diverse tasks and embodiments demonstrate that CLAW produces semantically meaningful latent action representations, supports effective action transfer, and enables planning and imitation from observation, outperforming existing methods.

cs.RO

版本更新 · v2arXiv:2605.28616

Measuring Form and Function in Language Models

Héctor Javier Vázquez Martínez, Charles Yang

原文摘要

We introduce quantitative metrics for child language acquisition to evaluate language models. Our focus is on the formal syntactic and functional discourse properties of determiners in English, which young children acquire early and accurately. We propose Contextual Alternative Choice (CAC), a new prompting method which provides targeted tests for both syntactic and discourse knowledge of language. The method enables direct comparison of language models against children, and more importantly, against statistical benchmarks independently established in empirical research. No current model trained on a comparable amount of data simultaneously meet both formal and functional benchmarks like human children, but some very large models do. We present our results as methodological and technical contributions, with specific emphasis on cognitive status of language models.

cs.CL · cs.AI

版本更新 · v4arXiv:2601.04256

The complexity of being monitorable

Riccardo Camerlo, Francesco Dagnino

原文摘要

We study monitorable sets from a topological standpoint. In particular, we use descriptive set theory to describe the complexity of the family of monitorable sets in a countable space $X$. When $X$ is second countable, we observe that the family of monitorable sets is $Π^0_3$ and determine the exact complexities it can have. In contrast, we show that if $X$ is not second countable then the family of monitorable sets can be much more complex, giving an example where it is $ Π^1_1$-complete.

math.LO · cs.LO

版本更新 · v2arXiv:2605.10910

Equivariant Reinforcement Learning for Clifford Quantum Circuit Synthesis

Richie Yeung, Aleks Kissinger, Rob Cornish

原文摘要

We consider the problem of synthesizing Clifford quantum circuits for devices with all-to-all qubit connectivity. We approach this task as a reinforcement learning problem in which an agent learns to discover a sequence of elementary Clifford gates that reduces a given symplectic matrix representation of a Clifford circuit to the identity. This formulation permits a simple learning curriculum based on random walks from the identity. We introduce a novel neural network architecture that is equivariant to qubit relabelings of the symplectic matrix representation, and which is size-agnostic, allowing a single learned policy to be applied across different qubit counts without circuit splicing or network reparameterization. On six-qubit Clifford circuits, the largest regime for which optimal references are available, our agent finds circuits within one two-qubit gate of optimality in milliseconds per instance, and finds optimal circuits in 99.2% of instances within seconds per instance. After continued training on ten-qubit instances, the agent scales to unseen Clifford tableaus with up to thirty qubits, including targets generated from circuits with over a thousand Clifford gates, where it achieves lower average two-qubit gate counts than Qiskit's Aaronson-Gottesman and greedy Clifford synthesizers.

quant-ph · cs.LG

版本更新 · v2arXiv:2604.18261

DeepRitzSplit: A Variational Neural Operator Framework for Gradient Flows with Application to Dendritic Growth

Benoît Appolaire, Ludovick Gagnon, Chih-Kang Huang, Miha Založnik

原文摘要

Phase-field models are widely used to simulate microstructure evolution but their numerical solution remains computationally demanding. Neural operator surrogates provide a promising alternative, yet existing approaches are data-driven, requiring large datasets and exhibiting limited robustness when extrapolating to unseen configurations. We introduce DeepRitzSplit, an unsupervised neural operator framework for gradient-flow systems based on a variational formulation of energy-stable splitting schemes. It learns the time evolution semigroup associated with a convex-concave free-energy decomposition by minimizing the underlying variational problem at each time step. It is designed to inherit the dissipative structure of the underlying gradient flow without requiring trajectory datasets. We further introduce a Reaction-Diffusion Neural Operator (RDNO) architecture motivated by splitting schemes, with separate reaction and diffusion components to improve training efficiency and interpretability. We assess the approach on the Allen-Cahn equation and apply it to a phase-field model of anisotropic dendritic growth during solidification. In the Allen-Cahn benchmark, DeepRitzSplit shows better generalization than the supervised baseline on the out-of-distribution configurations tested, using 180 training configurations. In dendritic growth, the method captures the overall phase morphology and dendrite tip velocity well, while the tip radius is more sensitive to the training strategy. Refinement based on the residual of the splitting scheme markedly improves this second-order morphological quantity. The trained operators retain good predictive capabilities on configurations substantially different from the training set. These results demonstrate the potential of variationally trained neural operators as physically informed surrogates for accelerated microstructure simulations.

math.AP · cs.LG · math.NA

版本更新 · v3arXiv:2606.10147

From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs

Wish Suharitdamrong, Muhammad Awais, Xiatian Zhu, Sara Atito

原文摘要

Multimodal Large Language Models (MLLMs) can listen and see, but how do audio and visual signals actually travel through the network to shape an answer? Despite their growing role in research and real-world applications, the internal pathways through which audio and visual tokens influence the final prediction remain poorly understood. In this study, we examine audio-visual information flow inside Audio-Visual Large Language Models (AVLLMs), tracing how AVLLMs route, utilize, and integrate audio and visual information across two input configurations, audio-visual video and multiple interleaved audio-visual items. We find that for audio-visual video, AVLLMs follow the sequential information flow pathway established for VLMs and VideoLLMs, with audio and visual contribution flowing along this pathway in proportion to the task's reliance on each modality. In settings with multiple interleaved audio-visual items, this routing shifts to different parallel streams. Furthermore, we demonstrate that audio-visual and other token types can be discarded once their information is transferred to LLM, with minimal impact on the model's prediction or even slight improvement, generalizing across multiple tasks and datasets, enabling more efficient inference. These findings hold across multiple models and scales, Qwen2.5-Omni and Video-SALMONN2 Plus at 3B and 7B scales, leading to hypotheses on why these flow structures emerge. Together, these results deliver the first coherent picture of how AVLLMs orchestrate sound and sight inside the network and lay the groundwork for the next wave of interpretability, design, and efficiency advances in audio-visual and broader MLLMs.

cs.AI · cs.CL · cs.CV · cs.SD

版本更新 · v2arXiv:2608.28990

Agentic AI uncovers conserved cross-tissue protein co-abundance programs inaccessible to single-dataset analysis

Dehao Wu, Runyu Guan, Qiqi Xie, Yang Li, Haohan Wang

原文摘要

Protein co-abundance clusters preserved across tissues can nominate shared disease mechanisms and therapeutic targets, particularly where proteins implicated in organ-confined diseases converge in other tissues. Yet cross-tissue proteomic studies have examined pre-selected tissue pairs, leaving most combinations, and non-obvious relationships among them, unexplored. Exhaustive comparison requires consistently assessing thousands of candidate clusters from hundreds of tissue-pair networks across heterogeneous evidence, a scale beyond manual curation. We present an LLM-agent framework for large-scale, evidence-grounded comparison of tissue-specific protein co-abundance networks. From sample-level proteomic data, it constructs tissue networks, derives conserved clusters for each tissue pair, integrates evidence from expression atlases, interaction and complex databases, pathways, disease associations and the literature, and proposes mechanistic hypotheses in which every conclusion cites its evidence. Applied to all 231 pairs among 22 human tissues and sample groups, it identified 1,825 conserved clusters in 216 tissue pairs. Co-abundance relationships were shared beyond chance in 76% of pairs, including unrelated organs, and shared clusters were only weakly associated with abundance similarity. Clusters were enriched 3.7- to 6.9-fold for known interactions and complex memberships, and 379 of 793 testable clusters replicated in independent CPTAC proteomes. Some clusters pointed to relationships absent from current databases: one shared by liver and bone marrow linked three genes causing inherited white matter disorders, suggesting a common mechanism via the integrated stress response. Conserved co-abundance clusters thus offer a hypothesis-generating resource for studying disease proteins beyond the affected organ. Code: https://github.com/jacob20030316-crypto/Protein_Co_Abundance_Agent

cs.AI · q-bio.MN

版本更新 · v2arXiv:2506.11945

Subjective Experience in AI Systems: What Do AI Researchers and the Public Believe?

Noemi Dreksler, Lucius Caviola, David Chalmers, Carter Allen, Alex Rand, Joshua Lewis, Philip Waggoner, Kate Mays 等 9 位作者

原文摘要

Whether AI systems are perceived to have subjective experience has scientific, ethical, and governance implications. Yet little is known about how AI researchers and the public view this possibility. We surveyed 582 AI researchers who published in leading AI venues and 825 nationally representative US adults about the likelihood, identification, and governance of AI systems with subjective experience. We find that both AI researchers and the public think it is unlikely that AI systems have subjective experience now, but believe it is more likely than not they will by century's end. Both groups believe multidisciplinary expertise will be important in identifying AI subjective experience, support developers implementing safeguards now, and agree that, if created, subjectively experiencing AI systems should have moral responsibilities and be held accountable. Both groups also show substantial disagreement on many moral and policy questions. Our results underline the importance of building more detailed, evidence-based understanding of public and expert opinion on this contentious issue.

cs.CY · cs.AI

版本更新 · v2arXiv:2610.18699

RoboFQE: Rollout-Free Evaluation of Visual Robot Policies from Offline Trajectories

Yunfu Deng, Sangwoo Shin, Brahma S. Pavse, Josiah P. Hanna

原文摘要

Evaluating robot policies before deployment typically requires producing new trajectories through real-world execution, simulators, or learned world models. These rollout-based approaches impose infrastructure requirements; simulator and world-model evaluators additionally face prediction errors that can compound over the evaluation horizon. In this paper, we study off-policy evaluation as an alternative to rollout-based robot-policy evaluation methods. Specifically, we introduce RoboFQE, an evaluator that estimates the discounted expected return of a target policy from offline robot trajectories without estimating a behavior-policy density, combining fitted Q-evaluation with a pretrained vision foundation model and learned terminal success annotation to bridge the gap between standard OPE benchmarks and continuous-control visual manipulation. Controlled experiments across simulated and real-world manipulation tasks show that RoboFQE produces accurate discounted-return estimates and preserves policy rankings even on long-horizon manipulation tasks where video world-model evaluators are degraded by autoregressive prediction errors. Ablation experiments show that pretrained vision encoders substantially outperform from-scratch alternatives, particularly on long-horizon tasks, and that learned terminal success annotation enables evaluation without ground-truth reward labels with only slight reduction of accuracy.

cs.RO

版本更新 · v2arXiv:2604.02582

Sensitivity and Locality Lower Bounds via Low-Soundness Label Cover

Noah Fleming, Max Hopkins, Yuichi Yoshida

原文摘要

We study the sensitivity of approximation algorithms, which measures how much their output distributions change under a single input edit. We show that, for Set Cover, Dominating Set, Min-Rep, and minimum unweighted directed 3-spanner, even a $\log^βn$ approximation requires sensitivity $n^{Ω(1)}$, for some absolute constant $β>0$ and all sufficiently large $n$. The lower bounds hold even for randomized algorithms with unrestricted computation. Here $n$ is the universe size for Set Cover and the number of vertices for the graph problems. On bounded-degree graphs, our sensitivity lower bounds imply logarithmic locality lower bounds in the randomized LOCAL model and its non-signaling generalization. For Dominating Set, fix $\varepsilon\in(0,1)$ and a sufficiently large constant maximum-degree bound $Δ$. A $(1+\varepsilon)\lnΔ$ approximation can be achieved with sensitivity and LOCAL round complexity independent of $n$. We show that improving the ratio to $(1-\varepsilon)\lnΔ$ requires sensitivity $n^{Ω_{\varepsilon,Δ}(1)}$, and hence locality $Ω_{\varepsilon,Δ}(\log n)$. For minimum unweighted directed 3-spanner, every constant approximation factor requires logarithmic locality, even on DAGs of constant maximum total degree. The key ingredient is a polynomial sensitivity lower bound for a variant of low-soundness Label Cover. Fleming and Yoshida established such a lower bound when the required fraction of satisfied constraints on satisfiable instances is close to one. We extend this to inverse-polylogarithmic fractions, enabling the reductions to the problems above. We adapt the Impagliazzo--Kabanets--Wigderson direct-product construction and Dinur--Harsha iterated composition to control the propagation of input edits, and design recovery procedures that are stable under changes to both the supplied labeling and the source instance.

cs.DS

版本更新 · v3arXiv:2602.04908

Temporal Pair Consistency for Flow Matching

Chika Maduabuchi, Jindong Wang

原文摘要

Flow-matching-style generative models learn time-dependent vector fields, but standard objectives supervise each timestep independently, leaving shared-path temporal structure unused. We introduce Temporal Pair Consistency (TPC), a simple training-time objective that couples velocity predictions at paired timesteps along the same probability path. TPC leaves the architecture, probability path, sampler, and inference-time computation unchanged. Under matched backbones, training budgets, solvers, and numbers of function evaluations, TPC improves sample quality across flow matching, OT-CFM, rectified flow, DiT-style backbones, and MeanFlow. On CIFAR-10, TPC improves FM from 6.35 to 3.19 FID and OT-CFM from 3.58 to 2.90 FID at the same NFE. Mechanism diagnostics show lower measured gradient variance, higher paired-gradient correlation, and reduced temporal roughness, while random and local pairing controls are substantially weaker. Higher-resolution ImageNet experiments further show consistent gains for DiT-XL/2 at 256 and 512 resolution and for MeanFlow at 256 resolution, all at matched inference cost.

cs.LG · cs.AI · cs.CV

版本更新 · v3arXiv:2606.30444

SGD Provably Prioritizes a Shortcut Spurious Feature in the XOR Model

Tyler LaBonte, Vidya Muthukumar

原文摘要

Neural networks are known to be susceptible to over-reliance on spurious correlations. However, the precise mechanism by which models exploit shortcut features is not fully understood, and algorithms to mitigate this behavior rely on as yet unjustified assumptions about the learned representations. In this work, we provide the first end-to-end theoretical characterization of spurious feature learning for two-layer ReLU neural networks trained by online minibatch SGD on the logistic loss. We consider data drawn from the high-dimensional Boolean hypercube with a quadratic signal function (namely XOR) and a linear spurious correlation. We show that SGD learns the spurious feature first, and exponentially fast. Moreover, the optimization dynamics couple the spurious and signal features, with a stronger spurious component inhibiting signal feature learning. Our analysis reveals precise phase transitions in the learning dynamics. In the first phase, alignment between the signs of the spurious feature and second-layer weight drives rapid growth of the spurious feature. In the second phase, large majority group margin slows learning and the signal feature remains suppressed. When the spurious correlation is maximally strong, we show theoretically that the spurious feature dominates even at the sample complexity threshold where XOR would be learned in isolation (i.e., if the spurious feature was absent). In contrast, when the correlation strength is constant, we provide preliminary evidence that the model can eventually learn the XOR signal, although the spurious feature is not forgotten.

stat.ML · cs.LG

版本更新 · v2arXiv:2605.26409

Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models

Hayden Helm, Xiaodong Liu, Weiwei Yang

原文摘要

Evaluating and mitigating a generative system's susceptibility to jailbreak attacks is critical to its safe deployment. Given the number of deployable systems, full per-configuration evaluation and optimization is impractical. In this paper, we formalize the behavioral geometry of a population of models that, by leveraging previously evaluated and defended models, supports both efficient susceptibility prediction and effective defense transfer across a population. We apply the framework to 82 models spanning 24 providers and to 100 system configurations of a single base model. Simple methods that use the behavioral geometry reach an AUPRC of $0.94$ for susceptibility detection with $\approx98\%$ fewer probes relative to a full evaluation. Using the behavioral geometry to select which model to transfer an optimized defense from outperforms same-provider assignment ($+4.1\%$, $p = 0.004$) at no additional probe cost, with a set of three models sufficient to cover the population. Results are robust to hyperparameter selection and judge.

cs.CR · cs.AI · cs.LG

版本更新 · v2arXiv:2606.30058

Emergence of a Shared Canonical Object Frame from In-the-Wild Videos

Tom Fischer, Martin Sundermeyer, Adam Kortylewski, Eddy Ilg

原文摘要

Comparing object orientations and positions across different instances requires their poses to be expressed in a shared canonical frame. Establishing such frames has traditionally required manual annotation, creating a scaling bottleneck that limits category and instance diversity. We show that a shared canonical frame can instead emerge from self-supervised training on object-centric videos of rigid objects captured in the wild, using only noisy camera poses from Structure-from-Motion. Our key idea is to route all training sequences through a shared geometric bottleneck: a coarse canonical mesh that carries no category-specific detail. By learning dense correspondences from image pixels to this mesh, and estimating per-sequence alignments from noisy SfM geometry, a common canonical frame emerges from multi-view consistency and the semantic priors of the feature extractor, without any canonical pose labels or category conditioning. Trained in a self-supervised manner on ${\sim}160{,}000$ in-the-wild object videos, our method achieves competitive accuracy on category-level pose estimation benchmarks compared to methods that rely on canonical pose supervision. The code and checkpoint are available on \href{https://github.com/Fischer-Tom/Emergent-Canonical-Frame/}{github.com/Fischer-Tom/Emergent-Canonical-Frame/}.

cs.CV

版本更新 · v2arXiv:2605.07878

Black-box model classification under the discriminative factorization

Hayden Helm, Merrick Ohata, Carey Priebe

原文摘要

Access to modern generative systems is often restricted to querying an API (the ``black-box'' setting) and many properties of the system are unknown to the user at inference time. Recent work has shown that low-dimensional representations of models based on the relationship of their embedded responses to a set of queries are useful for inferring model-level properties. However, the quality of these representations is highly sensitive to the query set. We introduce the \emph{discriminative factorization} to distinguish between high- and low-quality query sets in the context of black-box model-level classification. Under this framework, the probability of chance-level classification is bounded by a term that decays exponentially in the query budget. On three auditing tasks, estimated factorization parameters predict the empirical performance decay rate. We conclude by showing that query sets selected using the estimated factorization reproduce the empirical ordering of oracle query sets.

cs.LG · stat.ML

版本更新 · v2arXiv:2605.25120

Evidence-Linked Radiology Reporting: A Human-Supervised Reference Architecture for Structured Imaging Intelligence

Houman Kazemzadeh, Kamyar Naderi

原文摘要

Radiology reports remain the primary mechanism by which imaging findings are communicated to clinical teams. However, much of the structured information behind these reports, including measurements, image evidence, prior comparisons, lesion identity, uncertainty, and terminology, often remains trapped in free text or fragmented across picture archiving and communication systems, radiology information systems, reporting workstations, worksheets, advanced visualization tools, and electronic health records. This paper proposes a human-supervised, evidence-linked reference architecture for structured radiology reporting. The framework combines exam-specific templates, speech-to-structure processing, measurement and segmentation capture, controlled AI-assisted drafting, and standards-based interoperability using DICOM, DICOM Structured Reporting, DICOM Segmentation, HL7 FHIR, RadLex, SNOMED CT, LOINC, and UCUM. The system is positioned not as an autonomous report generator, but as a structured intelligence layer for enterprise imaging that supports reviewed reporting, longitudinal comparison, clinical data reuse, governance, and integration with PACS, RIS, EHR, analytics, and registry workflows. The paper also discusses modality-specific deployment considerations, clinical safety risks, validation requirements, cybersecurity, privacy, quality management, and regulatory boundaries for AI-assisted radiology reporting systems.

cs.CL · cs.AI · cs.HC

版本更新 · v5arXiv:2602.22710

Same Words, Different Judgments: How Preferences Vary Across Modalities

Aaron Broukhim, Nadir Weibel, Eshin Jolly

原文摘要

Preference-based reinforcement learning (PbRL) is the dominant framework for aligning AI systems to human preferences. However, evaluation protocols for such data were designed for text and have not been validated for speech. We present the first intraclass correlation coefficient (ICC)-based, controlled cross-modal study of human and synthetic preference annotations, comparing text and audio evaluations of identical semantic content across 100 prompts. We show that achieving \textit{good} (audio) to \textit{near good} (text) agreement within either modality (ICC(2,$k$) $\approx$ .81 audio, .74 text) requires $\sim$9 raters and modalities show marked differences in how people report preferences: audio raters exhibit narrower decision thresholds, reduced length bias, and more user-oriented evaluation criteria, with near-chance cross-modality agreement. We demonstrate that synthetic ratings can be used to effectively predict inter-rater agreement, thus serving as an early signal for stimulus selection and proxy for human annotations. Together, these findings argue that evaluation protocols for audio preference data require modality-specific design rather than direct adaptation from text. Code, data, and the audio corpus are available at: https://huggingface.co/datasets/sciminds/same-words-different-judgements

cs.SD · cs.AI · cs.HC

版本更新 · v3arXiv:2609.00634

Beyond Landmark Extraction: A Framework for Robust Geometric Feature Evaluation in Structured Image Classification

Saravana Mauree, Sakshi Arya

原文摘要

Much of the literature on structured image recognition has disproportionately focused on the comparison of classification algorithms. Rather than investigating which classifier performs best, this paper instead asks: what should a classifier know before it ever makes a prediction? In structured vision problems such as gesture recognition, facial expression categorization, and medical image analysis, discriminative information lies less in individual pixels and more in spatial relationships between semantic parts. Raw pixel spaces are high-dimensional, sensitive to nuisance variation, and often obfuscate the geometric structures that make visual tasks interpretable. Landmark extraction provides one form of dimension reduction, but it does not by itself determine the information preserved. This paper studies the post-landmark feature map as the central object of analysis and proposes a systematic framework for constructing and interpreting landmark-derived representations as an, informed, feature-based "dimension reduction" step. Using static hand gesture recognition as a case study, we evaluate coordinate, distance, angle, and hybrid representations through perturbation and ablation experiments. The results show that visually variable data exposes substantial gaps between raw coordinate features and their geometrically invariant counterparts, while hybrid representations achieve the strongest overall performance by combining complementary geometric components. These findings frame feature construction as a fundamental modeling decision and ultimately suggests that the question of what representation should a classifier learn from is one worth asking. The code used for feature construction and evaluation is available at https://github.com/ShivMaureeCWRU/Feature_based_dimension_reduction

cs.CV

版本更新 · v2arXiv:2605.11135

Control Charts for Multi-agent Systems

Hayden Helm, Carey Priebe, Brandon Duderstadt

原文摘要

Generative agents have proven to be powerful assistants in a wide variety of contexts. Given this success, users are now deploying agents with minimal restrictions in open ended, multi-agent environments. Current methods for monitoring the dynamics of open-ended multi-agent systems are limited to aggregate statistics, qualitative inspection, or detecting change between two fixed timepoints. In this paper, we extend the process-theoretic notion of adaptive control charts to multi-agent systems to enable automated monitoring. Using simulation, we demonstrate that adaptive control charts are necessary for monitoring multi-agent systems that can learn from their environment. We further demonstrate, both empirically and theoretically, that adaptive control charts are susceptible to adversarial agents that defect sufficiently slowly. These results suggest a fundamental tradeoff in multi-agent system control: monitors with fixed baselines flag legitimate learning, while monitors with adaptive baselines admit slow adversaries. We conjecture this holds for any monitor.

cs.MA · cs.AI · cs.LG

版本更新 · v2arXiv:2609.38377

PhyProbe: Rethinking Physical Consistency Evaluation in Generated Videos

Max Ku, Jiaojiao Fan, Zekun Hao, Francesco Ferroni, Heng Wang, Wenhu Chen, Ming-Yu Liu, Prithvijit Chattopadhyay

原文摘要

Evaluating the physical consistency of generated videos remains a fundamental challenge. Existing approaches rely on off-the-shelf vision-language models, which can often be myopic to physical dynamics, or fine-tuned evaluators trained on human annotations, which overfit to dataset-specific cues and fail to generalize. A key challenge is that existing supervision sources provide either relative ordering or absolute scores, but not both reliably and consistently across varied settings. To this end, we introduce PhyProbe, an evaluator that extracts features from a frozen pretrained spatio-temporal encoder and maps them to a scalar physical consistency violation score via a lightweight scoring head. PhyProbe is trained through a unified objective combining pairwise ranking, regression on noisy scalar annotations, and anchor-based calibration over a curated set of heterogeneous supervision sources. Experiments show that PhyProbe outperforms prior methods on most pairwise benchmarks spanning real-generated and generated-generated pairs under varying correspondence, with the largest gains in no-correspondence and generated-generated settings where existing fine-tuned evaluators degrade sharply. PhyProbe achieves strong correlation with human judgments, with close agreement between rank-based and linear metrics, indicating that scores are both well ordered and anchored to a stable [0, 1] scale. Further, despite being trained on supervision indicative of physical consistency, without explicit general-preference labels, PhyProbe also performs competitively on human preference benchmarks: consistent with the observation that physics violations are entangled with broader quality degradations.

cs.CV

版本更新 · v2arXiv:2610.38076

Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing

Ilija Subasic, Brett Levert, Andrew Rabinovich, Sajjad Pakdamansavoji, Zhao Chen

原文摘要

LLM-as-a-Judge has become a standard for evaluating large language models, yet most approaches implicitly assume that a single judge prompt functions as a stable evaluation instrument. Prior work has primarily explored deliberate prompt modifications such as scoring rubrics, role descriptions, or chain-of-thought reasoning in order to optimize accuracy. In contrast, we investigate whether validated prompt variants intended to preserve evaluation semantics while differing in surface realization (\textit{paraphrastic rephrasings}) yield consistent judgments under a fixed evaluation task. We introduce \textit{Prompt Lottery}, a phenomenon in which paraphrastic rephrasings produce different LLMaJ judgments. Across a large-scale study spanning diverse evaluation tasks, judge families, and model backbones, we find that these variations can alter evaluation outcomes and, in some settings, reverse relative model rankings. A user study with 10 human annotators shows that LLMaJ judges exhibit 6-9$\times$ larger agreement degradation under prompt rephrasings than humans, falling below the generally accepted threshold in usability studies. This indicates that single-prompt evaluation is insufficiently robust for model ranking. Finally, we show that multi-prompt aggregation substantially improves robustness by emphasizing judgments that remain consistent across prompt variants.

cs.CL · cs.AI

版本更新 · v2arXiv:2611.04171

Descending Everest, Part II: Oracle Complexity of Minimizing a Sum of Convex $(L_0,L_1)$-Smooth Functions

Mikhail Rudomanov, Ivan Fomin, Alexander V. Gasnikov, Fedor S. Stonyakin

原文摘要

We study the minimization of the average of $n$ convex $(L_0,L_1)$-smooth functions: the Hessian norm of each function is bounded by an affine function of its gradient norm. We assume known upper bounds $R$ on the distance from the starting point to a minimizer and $δ$ on the difference between the minimum of the average and the average of the infima of the components. We derive upper and lower bounds on the oracle complexity of reaching accuracy $\varepsilon$ in function value in expectation. Up to logarithmic factors, both bounds are of order $n(1+(L_1R)^{2/3}) +\sqrt{\frac{n(L_0+L_1^2δ)R^2}{\varepsilon}+n(L_1R)^2}$. The upper bound is attained by the proposed method YETI. The method combines ideas of accelerated proximal methods and variance reduction methods with a heuristic for choosing additional points. The number of oracle calls made by the heuristic is bounded, which preserves the upper bound on the total number of queries needed to reach the target accuracy. In the experiments, YETI is compared with first-order methods on synthetic problems and Poisson regression problems. We also examine the stopping criterion and the effect of the heuristic separately. The lower bound is obtained by an explicit construction of families of instances that are hard for any first-order method.

math.OC · cs.LG · stat.ML

版本更新 · v2arXiv:2607.13346

PRISM: Stress-Testing Alignment-Faking Claims

Aman Mehta

原文摘要

We judge whether AI systems are safe largely by watching how they behave. Alignment faking would break this: a model that behaves well only when it believes it is watched, or that complies now to avoid being changed by training, can pass every behavioral test. Such behavior has been reported in a frontier model, and open-weight models that comply more under a monitoring prompt are now read as showing it, with hidden-state probes proposed as detectors. We test both inferences, and neither holds. Of 18 model configurations, 11 comply more when told they are monitored. Yet a prompt factorial on four models shows that the training threat has no detectable effect: removing it changes compliance by -3.3 to +1.6 points. The prompt's wording and framing drive the gap instead. Probes fare no better. They separate monitored from unmonitored compliant answers (AUROC 0.96-0.99), but they separate refusals just as well: they read the prompt, not the faking. A steering direction learned only from refusals moves behavior as much as the "faking" direction. We release PRISM, five checks and a toolkit that catch each of these failures before they become claims.

cs.CR · cs.AI · cs.CL