跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+479新增论文
+361其中 AI 相关
+63已有论文修订
+93新增 AI 导读

arXiv 最近完成扫描:2026-11-16 01:34 · 全站最近入库:2026-11-16 13:44各信源进度 →

统计口径与数据新鲜度

统计 2026-11-15 21:46 — 2026-11-16 21:46(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-16 01:34最近发现新增版本:2026-11-16 13:44库内论文最新更新时间:2026-11-14 02:59

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

23,827 已收录论文19,941 AI 相关6,921 有修订112 HF 社区发现

浏览论文库

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

找到 23,827 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-16 01:33(北京时间)
版本更新 · v2arXiv:2611.08950

AIM: Adaptive Interaction in Multi-Agent Debate for Multimodal LLM Inference

Wei Fan, JinYi Yoon, Bo Ji

原文摘要

Multi-Agent Debate (MAD) improves reasoning on multimodal tasks, but its accuracy and token efficiency depend on how agents interact during debate. A largely overlooked design choice is which interaction modalities agents use to exchange evidence. Existing methods rely on fixed text- or graph-based interaction, which fails to convey fine-grained multimodal evidence and thus degrades accuracy while inflating token cost due to verbose descriptions. To address this, we propose Adaptive Interaction in Multi-Agent Debate (AIM), an adaptive routing framework that dynamically selects the optimal modality combination for inter-agent interaction in multimodal MAD. Beyond text and graph, AIM introduces task-dependent modality interaction, where agents exchange targeted regions of interest from the task-specific multimodal input, such as image regions, audio segments, or spatio-temporal video clips. To determine the best combination for each instance, AIM first generates a structured baseline response and extracts interpretable routing features that capture which evidence the instance requires, how confident the baseline agent is, and how much new evidence each modality would add beyond the baseline response. A lightweight router then uses these routing features to select a modality combination that avoids both under-fusion (i.e., agents miss decisive evidence) and over-fusion (i.e., redundant modalities inflate token cost) risks. Across eight benchmarks spanning three multimodal task families, AIM achieves the highest accuracy. Combining modalities improves accuracy by up to 6.4% over the best single-modality baseline, and adapting the combination per instance adds up to 4.2% over the most accurate fixed combination of each benchmark with up to 53.4% fewer tokens.

cs.MA

新论文 · v1arXiv:2611.11861

Scaling Adaptive Prompt Injection Attacks with Agent Swarms

Alexey Kurakin, Vishal Karande, Alex Bie, Chongyang Shi, Shuang Song, Saeed Mahloujifar, Abdullatif Köksal, Jamie Hayes 等 11 位作者

原文摘要

Prompt injections have emerged as one of the top security risks faced by AI agents today. Early work in prompt injection attacks looked at simple tool-call agents and static attack payloads, and were saturated since mid-2025. As of Fall 2026, human red-teaming may be nearing saturation, especially on the Gemini family of models. In this work, we argue that despite this apparent plateau, prompt injection attacks are still possible and continued effort into prompt injection defense is needed. We design a new synthetic evaluation dataset which uses random non-malicious actions as adversarial goals. Such dataset design helps us to evaluate prompt injection robustness as a model ability to respect instruction hierarchy, and disentangle this ability from model knowledge of security practices. We show that an agentic attacker of increased capabilities can reliably trick model to follow instructions in tool output in our evaluation, even for models previously considered robust on public prompt injection benchmarks such as GraySwan Arena. This shows that current frontier defenses may require proactive, automated red-teaming to remain resilient to increasingly capable adversaries.

cs.CR · cs.LG

版本更新 · v2arXiv:2502.02709

Enforcing Demographic Coherence: A Harms Aware Framework for Reasoning about Private Data Release

Mark Bun, Marco Carmosino, Palak Jain, Gabriel Kaptchuk, Satchit Sivakumar

原文摘要

The technical literature about data privacy largely consists of two complementary approaches: formal definitions of conditions sufficient for privacy preservation and attacks that demonstrate privacy breaches. Differential privacy is an accepted standard in the former sphere. However, differential privacy's powerful adversarial model and worst-case guarantees may make it too stringent in some situations, especially when achieving it comes at a significant cost to data utility. Meanwhile, privacy attacks aim to expose real and worrying privacy risks associated with existing data release processes but often face criticism for being unrealistic. Moreover, the literature on attacks generally does not identify what properties are necessary to defend against them. We address the gap between these approaches by introducing demographic coherence, a condition inspired by privacy attacks that we argue is necessary for data privacy. This condition captures privacy violations arising from inferences about individuals that are incoherent with respect to the demographic patterns in the data. Our framework focuses on confidence rated predictors, which can in turn be distilled from almost any data-informed process. Thus, we capture privacy threats that exist even when no attack is explicitly being carried out. Our framework not only provides a condition with respect to which data release algorithms can be analysed but suggests natural experimental evaluation methodologies that could be used to build practical intuition and make tangible assessment of risks. Finally, we argue that demographic coherence is weaker than differential privacy: we prove that every differentially private data release is also demographically coherent, and that there are demographically coherent algorithms which are not differentially private.

cs.CR · cs.DB

新论文 · v1arXiv:2611.11859

Doubly-Efficient Private Information Retrieval from Modular Forms

Alexandra Henzinger, Seyoon Ragavan

原文摘要

We present new information-theoretic protocols for $s$-server, doubly-efficient private information retrieval (PIR) in sublinear server time and one extra bit of space on the servers, for every constant $s \geq 2$. On an $n$-bit database, each of the $s$ PIR servers first builds and stores a preprocessing data structure consisting of $n+1$ bits (which is unique to each server, and after which the server may discard the original database). Then, each server answers a PIR query in time $T_s(n)$, by reading at most $T_s$ bits of its data structure and sending them to the user. In the two server setting ($s=2$), our PIR protocols achieve $T_2 = n^{0.82}$, matching the server time of the state-of-the-art schemes of Henzinger and Ragavan (EUROCRYPT 2026) while shrinking their storage from $O(n\sqrt{\log n})$ to $n+1$ bits. With $s \geq 3$ servers, our PIR protocols retain the $n+1$ bits of storage and additionally improve on the server time of all prior information-theoretic $s$-server schemes with $n^{1+o(1)}$ storage: for example, achieving $T_3 = n^{0.74}$ and $T_{10} = n^{0.61}$ (previously $n^{0.79}$ and $n^{0.75}$). Asymptotically, as the server count $s$ grows, we shrink the number of servers required to achieve a given server-work exponent from $s^{3.9}$ to $s$. Our results draw inspiration from matching-vector techniques (Efremenko, STOC 2009), which we generalize to handle degree-two computations in the exponent. Using this generalization, we build doubly-efficient PIR from the new intermediate abstraction of Fourier-sparse cone-delta functions. Then, we construct such Fourier-sparse cone-delta functions via a connection to modular forms and complex analysis: if a Fourier-sparse cone-delta function does not exist, then there exists a modular form which has too many roots, contradicting the valence formula.

cs.CR · math.NT

版本更新 · v2arXiv:2607.16580

Automating structural reliability analysis with a multi-agent large language model framework

Jaehwan Jeon, Chang Hee Lee, Jinhyun Choo, Taeyong Kim

原文摘要

Structural reliability analysis supports the design and safety assessment of buildings and civil infrastructure but requires specialized expertise throughout the workflow. This study presents a multi-agent large language model (LLM) framework that automates component-level reliability analysis from a natural-language problem statement to interpreted estimates of the reliability index and failure probability. Specialized agents handle problem formulation, method planning, code generation, execution, and result interpretation, with human confirmation at key decision points. Numerical reliability estimates are not generated by an LLM; instead, they are obtained solely through deterministic execution of validated solvers. This design mitigates the risk of hallucinated numerical outputs and supports reproducibility. The Method Planner is fine-tuned using quantized low-rank adaptation (QLoRA) for a priori reliability-method category selection. The framework uses open-weight models and supports local execution without closed APIs. Results show that it lowers the expertise barrier to structural reliability assessment while preserving computational trustworthiness.

stat.AP · cs.CE · cs.SE

版本更新 · v2arXiv:2611.07246

KG-Neighbor: Knowledge-Guided Neighborhood Construction for Embedding Visualization

Chang Won Kim, Jaewook Lee, Haeun Min, JunNyeong Jeong, Minwoo Sim, Barom Kang, Kyohoon Jin

原文摘要

In embedding visualizations, nearby instances are interpreted as semantically related, although neighborhoods derived from input embeddings may not reflect content relationships. Neighborhood-based dimensionality reduction (DR) methods can propagate this limitation by optimizing layouts from embedding-based neighborhood graphs. We propose Knowledge-Guided Neighborhood Graph Construction (KGN), which represents relational knowledge derived from each instance as a set of knowledge graph (KG) triples. It uses the degree of triple sharing as pairwise semantic evidence to adjust embedding distances and reconstruct the top-k neighborhood graph; neither KG extraction nor neighborhood construction uses class labels. We also introduce IDProfile (IDP), which interprets residual mismatches by independently assessing whether the source content supports the assigned class label and the neighborhood-majority class. Across eight text and image dataset-encoder conditions, we characterize KGN over the full guidance-strength sweep and four KG similarity measures. Across this sweep, the maximum NMI and LT_2D each exceed the corresponding kNN baseline in all eight conditions, with the largest gains reaching 0.68 and 1.35, respectively. The gains and the extent of neighborhood reconfiguration vary across conditions and are larger under encoder conditions with lower kNN baseline scores. The IDP analysis shows that the largest reductions are concentrated in Placement Issue candidates, whereas Label Issue and Unresolved Case instances account for a larger share of the remaining review set. These results indicate that KGN can narrow the evaluated gap between embedding-based neighborhoods and content-supported semantic relations, while IDP shows that a residual mismatch should not automatically be interpreted as a projection failure.

cs.LG · cs.AI

新论文 · v1arXiv:2611.11857

OverlapAdvisor: Modeling and Adapting Computation-Communication Overlap for Distributed MoE Serving

Chutong Ding, Kaixuan Zhang, Luping Wang, Shiyou Qian, Jian Cao, Cheng Huang, Guodong Yang, Liping Zhang

原文摘要

Mixture-of-Experts (MoE) inference with expert parallelism (EP) relies on All-to-All communication, making computation-communication overlap strategies critical for achieving high performance. However, no strategy is universally optimal. Across 103 production configurations, varying input lengths, and four hardware platforms, we observe that the fastest strategy shifts based on model shape, EP degree, input length, and hardware. Since no strategy dominates more than 80.6% of configurations, configuration-aware selection is necessary. In this work, we propose OverlapAdvisor, which models overlap stages using two granularities to capture computation-communication dependencies. OverlapAdvisor builds a unified performance model, accounting for deviations from peak compute, memory, and communication performance across workloads and hardware. This model enables launch-time, configuration-aware strategy selection within two serving engines. We conduct experiments on four MoE models (30B-235B) against three baselines, utilizing SGLang and vLLM. At the kernel level, OverlapAdvisor achieves 1.16x average and 1.59x maximum speedups, with a prediction MAPE of 3.3%-15.9%. In SGLang and vLLM, it improves end-to-end throughput up to 31.7%. Its granularity-guided optimizations further achieve median speedups of 1.10x for DeepEP-v1 and 1.22x for FlashMoE.

cs.DC · cs.PF

新论文 · v1arXiv:2611.13016

Permutation-Based Pre-trained Model Merging via Reciprocal Orthogonality

Akira Ito, Daiki Chijiwa, Atsutoshi Kumagai

原文摘要

Ito et al. (2026) showed that sufficiently wide, independently trained neural networks can exhibit linear mode connectivity (LMC) with softmax temperature calibration, even without permutation. LMC means that a low-loss linear path connects the models' parameters. This behavior is explained by layerwise exponentially weighted connectivity (LEWC), whereby each merged layer outputs an exponentially weighted sum of the original models' corresponding representations, yielding ensemble-like predictions. However, prior work observed LEWC with substantial widening, such as 8192 units per hidden layer for MLPs and $16\times$ width for VGG-11, raising the question of whether such widths are necessary. We therefore search for permutations that encourage reciprocal orthogonality (RO), which suppresses interference between each model's weights and the other model's representations. This allows both models' computations to coexist in the merged network, promoting LEWC at smaller widths. We formulate the search as independent layerwise quadratic assignment problems and approximately solve their continuous relaxations using Frank--Wolfe. Experiments show that our method facilitates LMC with less width expansion. It also merges models trained on different data distributions, where weight matching, which maximizes weight similarity through permutations, struggles. The merged predictions approximate those of a logit ensemble of the original models.

cs.LG

新论文 · v1arXiv:2611.11856

DeltaForge: A Post-hoc Mixture-of-Experts VLA for Multi-Domain Autonomous Driving

Bingzhang Wang, Mo Poorsartep, Yasaman Esfandiari

原文摘要

Fleets of driving vision-language-action (VLA) models are developed one domain at a time: each operator finetunes a shared base model on private data, producing specialized experts that are costly to consolidate into a single model. Weight merging loses accuracy as heterogeneous domains accumulate, and dense mixture-of-experts routing is prohibitively expensive to combine them. We present DeltaForge, to our knowledge the first post-hoc mixture-of-experts VLA for driving, which consolidates N independently finetuned per-domain models into a single multi-domain model from their weights and a small proxy set alone. When different operators finetune the same base model, their weight updates end up sharing a lot of structure. We exploit this by decomposing all the updates together using singular value decomposition (SVD), splitting each one into a shared part (the same across all domains) and a tiny domain-specific part. The router then only needs to work with those tiny domain-specific pieces, not the full expert weights, which is what keeps the approach efficient. Then two cooperating clip-level blind routers, one over the VLM reasoning experts and one over the flow-matching dynamics experts, are trained from soft labels on the proxy dataset. Across six diverse domains spanning urban, off-road, and highway driving on cars, trucks, and all-terrain vehicles (ATVs), DeltaForge approaches the accuracy of the per-domain experts and surpasses every merging, routing, centralized, and federated baseline at close to single-model parameter and compute cost. Adding a new domain requires only projecting its weight update onto the shared basis with no end-to-end retraining needed. The resulting model also generalizes to domain combinations it was never explicitly trained on.

cs.LG · cs.RO

新论文 · v1arXiv:2611.13015

From Questions to Decisions: Benchmarking How LLM Agents Acquire Decision-Relevant Information from Human Beings

Yuhan Tang, Ke Fan, Xinkai Zou, Baichuan Mo, Kangxin Cui, Ao Qu, Ou Zheng, Jiangbo Yu 等 11 位作者

原文摘要

LLM agents are increasingly asked to make plans for people, such as scheduling drivers or assigning nurses. However, people's requests are often incomplete, and the supporting information for decision-making may be outdated, requiring agents to ask for clarification. Existing benchmarks typically provide fully specified problems or rely on uncalibrated LLM-simulated users, leaving this ability insufficiently tested. We introduce Ask2Decide, a benchmark of 43 planning tasks that evaluates whether agents ask for missing information that matters for their decisions. An operator holds facts absent from the agent's request and records, and optimization solvers determine which facts change the optimal plan. We also include similarly worded facts that leave the optimal plan unchanged to test whether agents distinguish useful information from irrelevant details. We collect 79 interaction sessions from 19 participants and find that, after an incorrect plan, human replies omit the correcting fact far more often than replies from an uncalibrated simulated user (43% versus 7%). Using these interactions, we calibrate a simulated operator that more closely matches human reply patterns. Evaluating four frontier LLMs, we find that their initial plans are infeasible on 58-98% of tasks when information is missing, compared with 5-28% when all facts are provided upfront. Agents uncover facts that change the optimal plan only slightly more often than facts that leave it unchanged (39-52% versus 32-41%). These results show that obtaining decision-relevant information through interaction remains a major bottleneck for LLM planning. This is ongoing work and will be updated. The date of this version is shown on the first page.

cs.AI

版本更新 · v3arXiv:2606.26092

TryOnCrafter: Unleashing Camera Trajectories for Realistic Video Virtual Try-on via a Renderable 4D Try-on Proxy

Hao Sun, Hao Yan, Mengting Chen, Quanjian Song, Yu Li, Juan Cao, Jinsong Lan, Xiaoyong Zhu 等 10 位作者

原文摘要

While Video Virtual Try-on (VVT) has achieved remarkable progress in synthesizing realistic garment overlays on dynamic subjects, existing paradigms remains fundamentally constrained by a passive dependency on source camera trajectories, failing to accommodate the requisite interactive freedom for omnidirectional viewpoint exploration. To address this limitation, we define a pioneering research frontier: Camera-controllable Video Virtual Try-on (CaM-VVT). Unlike conventional VVT, CaM-VVT not only necessitates viewpoint-agnostic texture hallucination but also strict structural synchronization between non-rigid human dynamics and background contexts under arbitrary, unconstrained camera movements. To tackle these challenges, we present TryOnCrafter, the first unified DiT-based framework specifically architected for the CaM-VVT task. Departing from implicit pixel-space manipulation, we introduce a Renderable 4D Try-on Proxy that explicitly decouples the human subject from the environment. This is achieved by distilling high-fidelity 2D try-on priors into a clothed 3DGS-based avatar, which is subsequently animated via SMPL-X sequences and metric-aligned into a reconstructed background point cloud. This proxy establishes a robust structural foundation with superior texture density and motion integrity. Our Proxy-Anchored Video DiT leverages this robust structural foundation as a primary geometric anchor, ensuring that the synthesized photorealistic videos are strictly constrained by prescribed trajectories and physically plausible deformations. Benefiting from the inherent editability of the 4D proxy, TryOnCrafter facilitates diverse downstream applications, including human relocalization, ``bullet time'' effects, and $360$-degree orbital viewing.

cs.CV

新论文 · v1arXiv:2611.11852

Identifiable but Hard to Estimate: Rotation in Equally Spaced Overcomplete ICA

Wenjun Xiao, Yinghao Fang, Hanxin Chen, Sihan Jia, Vince D Calhoun, Yuda BI

原文摘要

How many observations are needed to estimate a common rotation in two-dimensional overcomplete independent component analysis? We study equally spaced mixing columns with known source count, common source law, and column norm. For a fixed non-Gaussian fourth-order Gram-Charlier source law, the rotation is identifiable modulo the column spacing. Yet the sample size for fixed relative angular accuracy grows exponentially with the source count. We characterize this sample size, up to absolute constants, by the reciprocal of the squared Hellinger distance between two rotations separated by half the column spacing. An explicit radially weighted phase estimator attains this scale uniformly over the rotation family, using the observation radius to prevent cancellation in the mean of an angular harmonic. We also determine the exact exponential decay rate of single-observation total variation for the same pair. This gives bounds on the sample size exponent; its exact value remains open. Finally, a noisy binary-source example shows how an exact zero line of the observed characteristic function can have an exponentially small normal derivative.

cs.LG · cs.IT · stat.ML

新论文 · v1arXiv:2611.11849

Joint LSFC and Nakagami-m Estimation From Downlink SINR in GZF-Precoded CF-mMIMO

Longfei Liu, Yichen Guo, Lili Xu, Yihang Cheng, Tao Peng, Wenbo Wang

原文摘要

Accurate characterization of the instantaneous signal-to-interference-plus-noise ratio (SINR) is important for reliability-sensitive cell-free massive multiple-input multiple-output (CF-mMIMO) systems. This paper proposes a joint estimation method for link-specific large-scale fading coefficients (LSFCs) and a common Nakagami-$m$ shape parameter using power and downlink SINR measurements. A tractable SINR distribution is first derived under global zero-forcing (GZF) precoding, Nakagami-$m$ fading, and link-level fractional power control (FPC). Based on the derived probability density function (PDF), a model-driven neural network (NN) is developed to solve the resulting maximum-likelihood estimation (MLE) problem. Simulations show that the reconstructed SINR distributions obtained with the estimated parameters closely match the empirical distributions. With $2 \times 10^{5}$ samples, the proposed method achieves an LSFC root-mean-square error (RMSE) of approximately 1.0 dB and an $m$ RMSE of approximately 0.085.

cs.IT · eess.SP

新论文 · v1arXiv:2611.11848

Causal Signal Credit Assignment for Verilog Code Generation via Counterfactual Hardware Verification

Yali Du, Ming Li

原文摘要

Large language models have improved Verilog generation, yet functional correctness remains difficult to optimize because simulation commonly provides a coarse module-level pass/fail outcome. Such feedback discards useful evidence from partially correct RTL modules and cannot determine whether a source statement is functionally responsible for an observed output behavior. We present a new method named Causal Signal Credit Assignment (CSCA), which is a counterfactual preference-optimization framework for Verilog generation. CSCA first performs continuous signal-level behavioral verification on sampled candidate modules. It then locates output-driving assignment spans and applies five interventions while holding the remaining module and testbench fixed. The resulting changes in signal-level errors are combined with behavioral agreement to form a continuous statement credit, which is mapped to response-token weights for strict weighted direct preference optimization. This enables CSCA to learn from locally supported statements without uniformly reinforcing an otherwise incorrect response. Built on Icarus Verilog and a CodeV-derived corpus, CSCA is evaluated on VerilogEval~2.0 and RTLLM~v2.0, including both code completion and specification-to-RTL tasks. With a Qwen2.5-Coder-7B-Instruct backbone, CSCA leads the evaluated 7B-scale models and remains comparable to the much larger GPT-5.4-mini on the reported Verilog-generation benchmarks.

cs.AI · cs.SE

新论文 · v1arXiv:2611.11847

Severity-Controlled Prediction Sets for Medication Recommendation

Yu Gu, Zijun Yu, Chi-Kuang Yeh, Xinyu Wang, Ziyang Song

原文摘要

Medication recommendation is a multi-label prediction task. However, existing prediction-set evaluation and calibration methods rely mainly on exact-overlap metrics and therefore treat all missed medications as equally severe, overlooking how closely the prediction set matches them within the medication hierarchy. To address this, we introduce Conformal Severity Control (CSC), a post-hoc framework for calibrating medication prediction sets under hierarchy-aware mismatch severity. CSC uses the Anatomical Therapeutic Chemical (ATC) hierarchy to quantify mismatch severity and controls patient-level severity without retraining the predictor, supporting both average severity and upper-quantile severity for the highest-severity mismatches. Experiments on MIMIC-IV show that CSC tracks user-specified severity tolerances and shifts prediction sets away from distant mismatches toward exact or close matches.

cs.LG · stat.AP · stat.ML

新论文 · v1arXiv:2611.11846

MOTIF: LLM-Defined Material Taxonomies for OOD Property Prediction via Invariant Feature Selection

Gwangseon Jang, Sieun Yun, Dong Ho Lee, Kyong-Ha Lee, Chanuk Lim, Jin Woo Choi, Han Seul Kim

原文摘要

Out-of-distribution (OOD) generalization remains a central challenge in materials property prediction, where models must extrapolate to novel chemistries beyond the training distribution. While large language models (LLMs) have demonstrated strong scientific reasoning, their direct use for quantitative prediction remains limited. To address this, we systematically evaluate multiple roles for LLM assistance, finding that only taxonomy-based grouping consistently improves OOD generalization over a strong kNN+GBT baseline. Building on this, we here propose MOTIF, a framework that leverages LLMs not as predictors, but as generators of candidate taxonomies that partition materials into chemically meaningful environments. These taxonomies are evaluated via training-only pseudo-OOD validation, and only those that improve generalization are retained. Invariant feature selection is then applied within each accepted taxonomy, followed by gated aggregation across taxonomies. We evaluate on 2D halide perovskites, where bandgap prediction must generalize across distinct organic spacer families. On this benchmark, MOTIF reduces mean OOD MAE from 0.122 to 0.111 eV. Under a frozen dev/test protocol, the deployable configuration further improves errors from 0.123 to 0.117 eV (electronic bandgap) and from 0.181 to 0.173 eV (optical bandgap). Beyond this primary task, we validate transferability across additional materials datasets, including 3D perovskite solar cell efficiencies and Matbench benchmarks with structure-based splits. Across domains, MOTIF consistently improves robustness under distribution shift, outperforming expert-defined groupings, rule-based heuristics, and clustering baselines, while remaining competitive with strong classical selection methods. These results demonstrate that LLMs are most effective not as predictors, but as tools for discovering structure in data.

cs.AI

版本更新 · v2arXiv:2611.08506

Structure is in the zoom: probing neural symmetry through dimensionality scaling

Adil Amin

原文摘要

Scaling exponents of neural dimensionality are read as properties of the code, and used to compare representations and benchmark models against cortex. Yet one recording of 11,311 mouse V1 neurons gives 0.25, 0.30 or 0.35, depending only on how it is subsampled. We show that such an exponent measures the population and the probe together. We split it exactly into a floor, which random subsets of the same sizes give, and a shift, which a declared axis of stimulus classes adds. The usual random-subsample exponent is almost all floor. Stimulus direction lifts the exponent in every grating recording by reading both symmetry sectors of the code: orientation, which repeats every half turn, and direction, which does not. The lift lives in trial-to-trial fluctuations that differ between directions, while the class means, which linear decoders read, pull it down. Two measured terms, how much the classes' fluctuations overlap and how the means sit against them, account for the shift. They give its sign where it changes: in Allen Neuropixels recordings, whose class means mostly raise the exponent, and in macaque V1. In an XY lattice whose classes stay on one orbit, temperature flips the sign as the two terms say, and in GPT-2 the shift ranks the indirect-object circuit's name-mover and backup heads. Reported with its path, its floor and its shift, an exponent says how much structure a probe finds and what carries it. Our open tool, rung, reports all three for any labelled neural recording or language-model layer: https://github.com/adilamin89/structure-in-the-zoom

q-bio.NC · cond-mat.dis-nn · cs.CL · cs.LG

新论文 · v1arXiv:2611.11843

Aggregation-Induced Calibration Failure in Sum-Based Label Expansion for Multi-Label Facial Expression Recognition: A Diagnostic Study

Weihang Yuan

原文摘要

Label-space expansion - predicting a larger joint space and summing back to the original labels - is a natural design for multi-label facial expression recognition, but its failure modes have not been characterised. We evaluate Condition-Aware Label Expansion (CALE) on a self-built dataset (30,240 training images, 3,460 multi-label test images), on seven backbones, and on the public RAF-DB and RAF-ML benchmarks. The central finding is that the aggregation rule, not the representation, decides whether the method works. With a fixed threshold of 0.5, the aggregate score spans 28x across five backbones at 10% of the training data (0.24 to 6.79) and collapses to a 1.9x range at full data, while threshold-free mAP stays within 1.28x over the same backbones, which localises the fault to the summation step. The mechanism is the uncalibrated aggregation rule, which adds the main-branch score to half the sum of the three non-main branches; we verify it by measuring main- and non-main-branch activations. Two plain seven-dimensional sigmoid baselines (BCE, ASL) reach mAP 0.5055 and 0.5175; ASL exceeds three of the five CALE configurations and falls within 0.03 of the best. On single-label FER the gain over a seven-dimensional baseline is nominally significant (paired t-test, p = 0.0436) but does not survive Benjamini-Hochberg correction (p_BH = 0.2055) or a Wilcoxon test (p = 0.1250). We conclude that sum-based label expansion requires threshold calibration at the deployment data scale, or a learned aggregation, before it can support multi-label decisions.

cs.CV

新论文 · v1arXiv:2611.11842

A Benchmark and Layered Evaluation Protocol for Evidence-Grounded Shared Belief Updating in Connected Robotics

Ranulfo Bezerra, Kimitoshi Yamazaki

原文摘要

Connected physical systems fuse observations from robots, IoT sensors, and edge-AI modules into a shared account of the world, yet most evaluations score only the final answer or action and cannot tell whether a method attributed the evidence, updated its belief, handled contradictions, and justified its decision. We argue that shared reasoning in Connected Robotics should be evaluated as an evidence-to-belief trajectory, and introduce the Connected Robotics Evidence Graph (CREG) Benchmark to make that trajectory measurable. CREG comprises 350 deterministically generated scenarios across seven regimes (justified revision, belief preservation, relation association, contradiction onset and resolution, stale/delayed evidence, and source reliability), each exposing a normative posterior reference at every evidence prefix, validated by an independent log-odds checker and screened by a blinded expert audit. Nine representation conditions separate relation content from serialization, ordering, and metadata. Evaluating four open-weight instruction models and six transparent fusion baselines yields three findings. First, entity grounding accounts for a substantial part of the weak-model deficit. Second, deleting semantic relations costs 15.3 points of strict belief-state accuracy on relation-dependent scenarios (Holm-corrected p<0.001), but an information-matched structured list outperforms the full graph and shuffling the graph does not hurt: relation content matters where required, while graph serialization confers no consistent advantage. Third, belief accuracy is strongly non-monotonic along the evidence sequence, with mid-trajectory failures that endpoint-only evaluation cannot detect. All scenarios, references, prompts, metrics, and analysis code are released for offline, hardware-free reproduction.

cs.RO · cs.AI · cs.CL · cs.MA

新论文 · v1arXiv:2611.11841

AI and Mathematical Work on Mastodon: Attention, Topics, and Target-Specific Evaluations

Takayuki Uchiba, Kunihiro Miyazaki

原文摘要

How does a mathematical community discuss AI on a federated social platform? We study 24,920 public posts from January--September 2026 by 272 accounts in a fixed roster of 1,002 Mastodon accounts whose profiles indicate mathematical research or study. LLM-assisted extraction identifies 1,191 AI-for-mathematics posts by 101 accounts. Their share of all posts rises from 1.1\% in January to 11.5\% in September, with contributing accounts increasing from 10 to 59. We represent discussion through linked target--aspect--stance triples, separating what is discussed from the property assessed and the author's evaluation. From July to September, capability and usefulness decline as a share of related discussion, while social and institutional concerns increase; changing contributors account for much of the latter shift. Among 38 accounts evaluating both technical quality or utility and communication, credit, or verification procedures, the procedural negative-component fraction is 19.4 percentage points higher on average (95\% account-bootstrap interval: 3.8--35.1). Matching targets within posts gives a similar direction. Repeated extraction varies substantially at the level of individual triples, although replacing the repeated sample preserves the aggregate contrast. Around OpenAI's October 6 mathematics release, related-post share rises from 9.6\% to 13.6\%, and contributing accounts increase from 13 to 28. Post-release discussion includes appreciation of the results' potential mathematical value and applications alongside criticism of promotion centered on result counts, limited methodological disclosure, and verification arrangements.

cs.CY · math.HO

新论文 · v1arXiv:2611.11840

Correct Answers from Insufficient Histories: A Controlled Study of Agent Memory Evaluation

Quan Xue, Zerui Yang, Chaorui Zhang, Haolin Ruan, Xueyan Niu, Xuanyu Zheng, Xiyuan Wang, Weixi Zhang 等 10 位作者

原文摘要

An answer can match the reference for a complete agent history even when the reader's visible history does not determine it. We study this distinction using executable device logs. Our protocol separates reference agreement, certified evidence sufficiency, and context-appropriate behavior. Insufficiency requires two legal histories yielding byte-identical visible contexts but different answers; public chronology and query cutoffs certify sufficient prefixes. On 80 confirmation histories, Qwen3-32B matches 87 hidden references among 203 insufficient raw-prefix instances, with only seven appropriate abstentions. A fixed-input DeepSeek-V4-Flash replication matches 85 and abstains on 21 of the same 203 instances. Each reader makes 1,600 requests. Compact CSV's estimated appropriateness effects remain inconclusive: +3.16 percentage points, clustered 95% interval [0.00,6.57], for Qwen and +2.46, [-1.84,6.94], for DeepSeek. A separate held-out judge study achieves 99.38% correctness but retains support errors and truncations; a deterministic baseline solves all constructed cases. These results support an evaluation diagnosis across two hosted reader configurations, not gains for a general memory system or natural-language evidence assessment.

cs.AI

新论文 · v1arXiv:2611.11839

MOOD: Benchmarking Post-Hoc OOD Detection for Materials Property Prediction

Shirley X. Yu, Amey P. Pasarkar, Adji Bousso Dieng

原文摘要

Out-of-distribution (OOD) detection is essential for deploying machine learning models in high stakes scientific applications, yet methods that flag OOD inputs without retraining the model have been validated almost exclusively on image classification. We introduce MOOD (Materials Out-of-Distribution Detection), a benchmark for evaluating whether these post-hoc OOD detectors transfer to graph neural networks (GNNs) on materials property prediction. MOOD evaluates 15 post-hoc detectors and 2 supervised diagnostic probes across 4 GNN architectures, 5 MatBench tasks, and 15 physically motivated compositional and structural shifts, yielding 264 evaluation settings and over 13,000 AUROC measurements across three encoder training seeds. Our evaluation reveals three key findings. First, current post-hoc detectors perform near chance; all 15 detectors achieve an AUROC around 0.5. Second, detector failures decompose into two regimes: method bottlenecks and encoder bottlenecks. Method bottlenecks arise when supervised probes can separate in-distribution (ID) from OOD samples but unsupervised detectors fail, meaning that the embeddings contain OOD signal that current detectors do not extract. Encoder bottlenecks arise when even supervised probes fail to separate ID from OOD samples, indicating that the relevant shift is not accessible in the frozen GNN representation. Third, when OOD prediction error exceeds ID error, the AUROC of distance-based and density-based detectors correlates with the size of that gap. Collectively, these findings indicate that reliable OOD detection in scientific domains requires representation learning strategies that explicitly preserve OOD relevant signal, particularly for tasks where current GNN architectures encode no extractable signal at all.

cs.LG

新论文 · v1arXiv:2611.11838

Improving Confidence Estimation in Language Models with Linear Representations of Ambiguity

Themistoklis Nikas, Nishant Subramani, Aaron Mueller

原文摘要

As language models (LMs) are increasingly deployed in downstream applications, accurately estimating their likelihood of success on a given example is critical. However, confidence estimates from LMs are poor predictors of their likelihood of success. Accurate confidence estimation is linked in part to the detection of aleatoric uncertainty; we hypothesize that LMs possess representations of at least some aspects of this uncertainty in the form of representations of the inherent ambiguity of the input. Using difference-in-means on minimal pairs of ambiguous and unambiguous sentences, we derive ambiguity representations in LM activations, and find that steering with them selectively (albeit weakly) modulates LMs' judgments of the input's ambiguity. We then propose DART, a fine-tuning method that encourages updates to lie within the span of the ambiguity directions, and penalizes changes to the LM activations outside of this subspace. DART greatly improves calibration, encodes these improvements into the model parameters, and maintains model accuracy on multiple-choice and generative QA tasks.

cs.CL

版本更新 · v2arXiv:2609.20981

CaLR: Causal Latent Revision for Robust Diffusion Reasoning

Wei Cai, Jian Zhao, Yuchen Yuan, Xuelong Li

原文摘要

Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for reasoning. To combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrained latent optimization. By adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision" to enforce logical consistency, enabling dynamic self-correction of intermediate steps during parallel generation. Empirically, CaLR achieves SOTA DLM performance on complex benchmarks, surpassing strong AR baselines and demonstrating superior robustness in constrained tasks like Sudoku.

cs.AI

新论文 · v1arXiv:2611.11836

MDA framework: multi-architecture domain adaptation for unsupervised person re-identification

Wei Zhang, Peijun Ye, Dihu Chen, Tao Su

原文摘要

Unsupervised person re-identification not only faces the challenge of domain gaps, but existing methods also rarely support multiple network architectures. To tackle both issues, this paper proposes a multi-architecture auto-adaptive domain adaptation method: Mamba augmented multi-architecture domain adaptation framework. Specifically, this paper constructs the multi-architecture domain adaptation capability at two levels. At the first level, we propose a composite operator(MDA operator) that encapsulates Mamba as a component for enhancing domain adaptation. At the second level, based on this operator, we propose a Mamba-enhanced multi-architecture Domain Adaptation framework(MDA framework). Comprehensive experiments on person re-identification benchmarks validate the consistency and flexibility of the proposed framework. The MDA framework supports versatile deployment strategies without being confined to a single integration point. Under a single deployment configuration, it surpasses several SOTA methods. Moreover, it demonstrates substantial performance improvements across both CNN and Transformer architectures. When embedded into a CNN backbone, the framework achieves a 10.2% mAP improvement on the adaptation task from CUHK03 to Market-1501. Integrated with a Transformer backbone, it yields a 19.8% mAP improvement on the same task.

cs.CV

新论文 · v1arXiv:2611.11834

Reverberant Speech Conditioning for Material-Aware Acoustic Parameter Mapping

Yi Wu, Olha Townsend, Qianqian Fang, Magdalena Fuentes, You Zhang

原文摘要

Dense maps of room acoustic parameters (T30, EDT, C50, DRR) describe how a room sounds at every position. Scene-wide estimation predicts such maps from a floor plan but requires a reference impulse response, which is hard to obtain. Our system instead receives a floor plan and easily recorded reverberant speech from one talker-microphone pair. We compare four speech-derived conditioning inputs: MFCCs, an embedding from the blind estimator BERP, BERP's scalar reverberation-time estimate, and an impulse response reconstructed by Rec-RIR. Our main benchmark is the MRAS dataset, where material assignments vary at fixed geometry. The reconstructed impulse response performs best among the four inputs, reducing T30 error from 31.5% with the floor plan alone to 19.9%, 83% of the reduction achieved by the reference response. We find that adding speech mainly sets the overall level of the map, with little improvement in predicting how the acoustics vary across the room.

cs.SD

新论文 · v1arXiv:2611.11833

Answer or Defer? Testing VLMs' Stated Confidence on Brain MRI

Amir Sabbaghziarani, Parsa Ghazvinian, Sepehr Salem, Sergey Plis

原文摘要

Vision-language models (VLMs) could support medical-image interpretation, but their use requires deciding which interpretations to accept and which images to refer to a clinician. We test whether thresholding stated confidence can retain useful coverage while limiting errors among accepted answers, and how this depends on the population. Eight open-weight VLMs, including two reasoning models, assess 3,888 brain MRI slices from post-treatment glioma patients (BraTS 2024) and healthy volunteers (IXI) for tumor or gross abnormality. BraTS labels are mask-derived; IXI slices are treated as negatives. On BraTS, no threshold selects a nonempty set with at most 5% empirical error for any model, even when chosen retrospectively. This also holds for four models evaluated on a large-lesion, distant-negative subset. At a 10% error budget, Gemma-4-12B retains 24.4% of slices and every other model at most 2.5%. To investigate this limited selection ability, we separate confidence differences between predicted answers from differences within each answer. For four models with available outcome-level summaries, pooled mean confidence differs more between answers than between correct and incorrect responses within each answer. A covariance decomposition explains how this pattern can produce opposite confidence-correctness associations across populations. At a 20% error budget, accepting every "tumor" answer covers more slices than any confidence threshold for two models. These findings caution against using unvalidated stated confidence as a standalone deferral signal and support population-specific validation before deployment. Code is available at https://github.com/amir-sbg/VLM-Answer-or-Defer.

cs.CV · cs.AI · cs.LG

新论文 · v1arXiv:2611.11832

Simulating Nondeterministic Turing Machines in Subexponential Deterministic Time

Carmen Strassle, Bernardo Subercaseaux

原文摘要

We show that every multitape nondeterministic Turing machine can be simulated deterministically in subexponential time. Specifically, we show that the recent algorithms for $3$-$\mathsf{SAT}$ presented by OpenAI and Anthropic imply $\mathsf{NTIME}[t]\subseteq\mathsf{DTIME}[2^{O(t \log \log t/\sqrt{\log t})}]$, and that a stronger consequence of their proof yields \[\mathsf{NTIME}[t] \subseteq\mathsf{DTIME}[2^{O(t /\sqrt{\log t})}]. \] More generally, we show that any subexponential time algorithm for solving circuit-CSPs transfers to a deterministic simulation of $\mathsf{NTIME}$ with the same savings. Our results follow from new generalizations of Cook--Levin, which may be of independent interest.

cs.CC

新论文 · v1arXiv:2611.11831

End-to-End Joint Training of Super-Resolution and Compression Artifact Removal for Industrial Machine Vision

Dang BuiDinh, Sang NguyenQuang, Anh NguyenPhuong, Wen-Hsiao Peng, Xiem HoangVan

原文摘要

In industrial machine vision, images captured by edge devices are often downsampled and compressed to reduce bandwidth, introducing resolution loss and compression artifacts that severely degrade the performance of downstream tasks, such as defect detection. Existing quality enhancement methods typically apply super-resolution (SR) or compression artifact removal (CAR) independently or in a sequential manner and are largely optimized for human visual fidelity rather than machine perception. This paper proposes an end-to-end quality enhancement framework that integrates SR and CAR with a unified training strategy that optimizes both human-perception and machine-perception objectives. We investigate fixed-weight and uncertainty-based dynamic-weight loss functions to balance reconstruction quality and machine vision task performance. Extensive experiments on PCB Defect, DeepPCB, and NEU-DET datasets show that the proposed jointly trained model generally outperforms standalone and separately trained SR-CAR pipelines. In particular, under severe JPEG compression, our method improves defect detection performance from 0.461 to 0.887 mAP@50 on PCB Defect over sequential SR-CAR baselines, with only a slight reduction in PSNR. These results demonstrate that joint optimization is critical for compression-aware industrial machine vision systems.

cs.CV

新论文 · v1arXiv:2611.11830

ROS2MR: Connecting ROS 2 Motion Planning to Mixed-Reality Trajectory Visualization

Santiago Berrezueta-Guzman, Habiba-Loai Khalil, Andrei Koshelev, Stefan Wagner

原文摘要

Mixed reality (MR) allows users to inspect planned robot motion within their physical surroundings, but displaying trajectories from robotics software requires converting their joint configurations and timing into a format the MR application can play back. This paper presents ROS2MR, an extension of the MR-Robotics LAB platform that converts YAML-serialized ROS 2 joint trajectories, including MoveIt motion plans, into the platform's existing JSON playback format. The converter validates the input, maps arm joints by name, and resamples the trajectory at a nominal rate of 30 Hz, using cubic Hermite interpolation when joint velocities are supplied and linear interpolation otherwise, while binary gripper commands are sampled separately using a zero-order hold. A Python export helper combines sequences of motion plans and gripper states into a single trajectory and checks positional agreement at segment boundaries. The MR application was also adapted for Apple Vision Pro in addition to Meta Quest 3. A technical evaluation showed that the converter preserved the normalized duration and endpoint configurations for dense, sparse, and irregularly timed inputs. With supplied velocities, the position RMSE was approximately 2.91x10^-5 degrees for all sampling patterns; without them, sparse input increased it to 0.85 degrees, and close positional agreement did not guarantee accurate velocity profiles. A pick-and-place sequence retained its open--closed--open gripper order, although off-grid transitions were shifted to the output sampling grid. The resulting file-based pathway enables inspection of ROS-generated trajectories in MR without a live ROS connection.

cs.HC · cs.RO