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

+1112新增论文
+875其中 AI 相关
+133已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-19 19:10 · 全站最近入库:2026-11-19 19:09各信源进度 →

统计口径与数据新鲜度

统计 2026-11-18 20:00 — 2026-11-19 20:00(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:2026-11-19 19:10最近发现新增版本:2026-11-19 19:09库内论文最新更新时间:2026-11-19 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。

27,982 已收录论文23,249 AI 相关8,559 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 20,663 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-19 19:07(北京时间)
版本更新 · v2arXiv:2609.35186

Elections, Fractional Dichromatic Number and Domination in Tournaments and Digraphs

Paul Colinot, Alantha Newman

原文摘要

The dichromatic number of a digraph is the minimum number of induced acyclic sets into which its vertex set can be partitioned. The fractional dichromatic number is its fractional relaxation. Bourneuf, Charbit and Thomassé showed that the domination number of a tournament can be bounded as an (exponential) function of its fractional dichromatic number. This naturally leads to the question of finding a polynomial bound. We show that this bound is linear. To do this, we show a connection between undominated sets in elections and dominating sets in tournaments with bounded fractional dichromatic number. Within this framework, we also connect the well-known question of the existence of small Condorcet winning sets in elections to domination problems in digraphs with fractional dichromatic number less than 2. Thus, we open a new path to investigating these popular open problems in social choice from the perspective of graph theory.

math.CO · cs.DM

新论文 · v1arXiv:2611.21216

MiSTiQ: MicroScaling Trellis Quantization for Extreme KV Cache Compression

Ofir Gordon, Lior Dikstein, Paul Balanca, Idan Achituve, Douglas Orr, Hai Victor Habi

原文摘要

Modern large language models (LLMs) increasingly support longer contexts and serve more concurrent requests, increasing both sequence lengths and batch sizes. Both trends increase KV-cache size and memory traffic during inference. KV-cache quantization reduces these costs, with recent approaches using codebook-based vector quantization to achieve extremely low bit widths. However, hardware constraints limit codebook size, and most methods reconstruct the cache in higher-precision floating-point formats for attention, limiting the computational benefits of quantization. To address these limitations, we introduce MiSTiQ, a drop-in replacement for codebook-based vector quantizers that combines trellis-coded storage with microscaling (MX) formats for low-precision computation. We combine trellis quantization with beam search to encode the KV cache over large reconstruction spaces without large codebooks. To support MX computation, we derive MX-Hash functions that generate reconstruction values directly in the target MX element format. These functions use the scale-conditioned marginal distribution of block-normalized Gaussian inputs. On Llama3 and Qwen3, MiSTiQ achieves lower WikiText perplexity and higher GSM8K accuracy than the evaluated baselines across several block-scale formats at approximately 1.25 effective bits per element. With NVFP4, it also achieves the highest LongBench average on both models.

cs.LG

新论文 · v1arXiv:2611.21215

Open-fMIND: A Large-Scale Open-Access fMRI Corpus for Pretraining Brain Foundation Models

Carlo Alberto Barbano, Giovanni Marraffini, Demian Wassermann

原文摘要

Self-supervised pretraining has transformed language and vision, but its application to functional MRI (fMRI) is constrained by data access. The largest standardized fMRI resources are governed by data use agreements and dominated by a small number of acquisition protocols, while the thousands of openly licensed datasets on OpenNeuro are individually small, heterogeneously organized, and unprocessed. We present the first release of Open-fMIND, a uniformly preprocessed, openly accessible corpus of parcellated fMRI timeseries from 322 OpenNeuro studies, comprising 11,379 subjects and 45,731 functional runs. Each run is preprocessed with fMRIPrep and parcellated under 52 atlases spanning 14 families, yielding a massive 2.4 million timeseries. Timeseries are released either denoised with our default pipeline (preproc) or as the fMRIPrep output with its confound files (quasiraw), with functional connectivity matrices, per-run provenance and motion metadata. Open-fMIND is intended as a pretraining resource for self-supervised brain representation learning. Pretraining on it yields downstream performance comparable to that obtained with UK Biobank.

cs.LG · q-bio.NC

新论文 · v1arXiv:2611.21214

Taking Time Out of TOCTOU: From Race Reproduction to Security Discovery

Hasnain Naeem, Ryan Henry

原文摘要

Time-of-check-to-time-of-use (TOCTOU) races are hard to reproduce and lost races can make vulnerable programs look fixed. We present Chronos, a framework that separates intervention timing from verdict evidence and builds reusable one-trial security regression oracles from declared races for deterministic reproduction, commit bisection and CI. Race Specifications compose reusable hold engines, actions, observers and outcome rules. When the declared hold is reached and the attack completes while held, Chronos orders it before the use under blocking and interference assumptions. Chronos uses seccomp notifications to enforce declared hold-attack-release ordering at syscall boundaries on stock Linux; ptrace and a KVM guest gate provide holds at userspace and kernel instruction sites. One decision procedure labels every mechanism's runs: fixed requires positive, case-specific evidence of preservation, justified refusal, omission under checked coverage, or a declared rule; missing or contradictory required evidence remains indeterminate. On known-vulnerable builds, effect-only checkers made over 800 false-fixed classifications in failed-intervention trials; Chronos made none. All 17 syscall-held cases and four of six instruction-held cases had vulnerable and fixed labels. Two kernel cases remain one-sided. All 27 historical bisections recovered the expected boundaries. For runc's masked-path bisection with one victim per launch, fixed retry and Git Bayesect took medians of 475 and 398 natural launches; Chronos used seven held trials including endpoint checks. Since a label means the same however a fix is written, we also ran the specifications and attack variants on releases, distribution packages and other runtimes. There, Chronos helped us uncover three issues: incomplete fixes in chrony and rpm, and a Debian rpm package its tracker listed as fixed.

cs.CR · cs.OS · cs.SE

版本更新 · v8arXiv:2604.11581

Hidden Measurement Error in LLM Pipelines Distorts Annotation, Evaluation, and Benchmarking

Solomon Messing

原文摘要

LLM evaluations drive model deployment, safety standards, research conclusions, and future projections, for example, about AI's impact on the labor market. Yet the way we often estimate noise from autoraters (LLM-as-judge) ignores variability from factors like judge model choice and prompt phrasing, which means we're underestimating error in a way that worsens with more data. This unmeasured variance can affect results enough to reverse conclusions. And judge-pipelines that fail to average this variation are more vulnerable to "benchmark hacking". This paper introduces a framework we call "total evaluation error (TEE)," which decomposes LLM pipeline uncertainty and uses design-study projections to reduce total error. In our safety and ideology applications, naive standard errors are 53-70% smaller than the corrected SE under our TEE framework. Using Chatbot Arena data, we show naive 95% CI coverage drops as n grows, while TEE-corrected coverage holds at 95%, and TEE-guided pipelines restrict the potential range of benchmark gaming by roughly 43%. We show further that a small pilot allows for the estimation of proper CIs, with the added benefit of projecting which design changes most improve precision. Using those projections to allocate API calls cut MMLU estimation error roughly in half against the answer key at the same budget, and increased per-match agreement with human voters on Chatbot Arena by 5.5 percentage points.

cs.CL

新论文 · v1arXiv:2611.21211

Teaching Language Models to Adapt Answer Granularity to Their Confidence

Lukas Ellinger, Georg Groh

原文摘要

When precise details are uncertain, a coarser answer can retain useful information without guessing or abstaining. We investigate whether language models can learn this behavior through model-conditioned supervision derived from consistency across sampled responses and recognition of answers at different granularity levels. We evaluate LFM2.5-1.2B, Ministral~3~3B, Qwen3.5-9B, and Gemma~4~31B on temporal and geographic questions and three generalization datasets. In several in-domain settings, fine-tuning improves the balance of correctness and informativeness beyond coarsening base answers to the best fixed level. Across generalization datasets, all four models produce coarser answers more often at low than high base-model confidence. The three smaller models' answers are also preferred over base responses on held-out relations when correctness and granularity are considered jointly. These findings demonstrate learnable confidence-related granularity adaptation, with benefits varying by model.

cs.CL · cs.LG

版本更新 · v5arXiv:2510.27408

Estimation of aboveground biomass in a tropical dry forest: An intercomparison of airborne, unmanned, and space laser scanning

Nelson Mattié, Arturo Sanchez-Azofeifa, Pablo Crespo-Peremarch, Juan-Ygnacio López-Hernández

原文摘要

Accurate aboveground biomass (AGB) estimates are needed to report forest carbon under the Paris Agreement, yet tropical dry forests (TDFs) remain among the least studied tropical ecosystems. We present a plot-level intercomparison of airborne (ALS), unmanned (ULS) and simulated spaceborne full-waveform (SLS, GEDI simulator) laser scanning for AGB estimation in ten permanent plots of Santa Rosa National Park, Costa Rica, covering early to late succession. Reference AGB (22-220 Mg/ha) was computed from the 2021 census with a documented procedure that screens diameter time series and propagates measurement, wood-density and allometric errors (plot uncertainty 3.5-14.5%). We compared ordinary least squares (OLS), a log-log model of mean canopy height fixed a priori and a single-metric model selected from the data, with support vector machine regression (SVM). Predictor selection, scaling and hyper-parameter tuning were nested within leave-one-plot-out validation. ALS gave the lowest errors: the a priori OLS model reached an RMSE of 32.3 Mg/ha (29%, R2_LOO = 0.77) and SVM 35.8 Mg/ha (32%). SVM did not improve on OLS for any platform and was significantly worse with ULS (p = 0.01). The best ULS and SLS models reached 33% and 37%; differences between platforms were not significant with ten plots. Selecting predictors on all plots before cross-validation lowered the apparent SVM error by up to 23 percentage points, and footprint coordinates, ALS point density and terrain elevation from the GEDI simulator produced spurious skill; these effects explain the higher accuracies reported in an earlier version of this work. Model errors remained at least 2.5 times larger than the field-AGB uncertainty. With small plot networks, a height-based OLS model with ALS is the most robust choice, and more and larger plots are needed before nonlinear models can be shown to add value in TDFs.

eess.SP · cs.LG

新论文 · v1arXiv:2611.21210

Full-Cover and Non-Overlapping Uni-domination in Graphs: Bounds and Complexity

Florent Foucaud, Yan Gerard, Christian Laforest, Le Vu Pham

原文摘要

A vertex is uni-dominated by a set $S \subseteq V(G)$ if exactly one vertex of $S$ lies in its closed neighborhood, and the uni-domination number $\mathrm{ud}^*(G)$ is the largest number of vertices a single set can uni-dominate. The graphs admitting an efficient dominating set are those with $\mathrm{ud}^*(G) = n$. Hedetniemi and Rall asked for the largest constant $c$ with $\mathrm{ud}^*(T) \geq c \cdot n$ for every tree $T$ of order $n$. We answer their question: the constant is $4/5$, and a set attaining it can be computed in linear time. This set is moreover a dominating set. Recall that an efficient dominating set satisfies two requirements at once: it dominates every vertex, and it dominates no vertex twice. Our set satisfies the first one, which leads us to study $\mathrm{ud}^*$ under each of the two requirements separately, giving the Full-Cover variant $\mathrm{ud}^*_0$ and the Non-Overlapping variant $\mathrm{ud}^*_+$. The two behave very differently: on arbitrarily large graphs $\mathrm{ud}^*_0$ may remain bounded by a constant, whereas $\mathrm{ud}^*_+$ always grows with the order. We present bounds for both, and show that the associated decision problems are polynomial for constant thresholds and NP-complete for the thresholds $\lceil \sqrt[c]{n} \rceil$ and $4\sqrt{n}$, respectively, the latter even on split graphs. Finally, we show that neither of the two parameters bounds the other.

cs.DM · math.CO

新论文 · v1arXiv:2611.21209

BoBDeLaN: Bounded-Bandwidth Context Adaptation for Deep Lagrangian Robot Control

Lucas Schulze, Sucheth Shenoy, Jan Peters, Oleg Arenz

原文摘要

Learned physics-based models provide structure for robot control, but uncertain payloads and unmodeled dynamics limit their predictive accuracy. Context-aware Deep Lagrangian Networks enable adaptation through a latent context inferred from measurement histories while retaining Lagrangian structure for each fixed context. However, this structure alone does not constrain sensitivity to measurement disturbances. We introduce BoBDeLaN, which combines a contracting, gain-bounded Recurrent Equilibrium Network context encoder with stable linear filtering to shape disturbance sensitivity across frequencies. This architecture guarantees exponential forgetting of the recurrent initialization and admits a finite-horizon, frequency-weighted bound on context perturbations, which extends to residual torque prediction under a Lipschitz continuity assumption on the context-conditioned dynamics model. We evaluate a low-pass realization within model predictive control on simulated and real hardware manipulation tasks, together with frequency-sensitivity analysis on real-robot data. Compared with an LSTM encoder using the same input filter, BoBDeLaN exhibits reduced high-frequency context sensitivity, lower residual torque prediction error, and smoother motion, with less degradation under injected disturbances.

cs.RO · cs.LG · eess.SY

新论文 · v1arXiv:2611.21208

Validating LLM-Based Essay Scorers: A Comparative Analysis in Critical Online Reasoning in Medical Education

Daniel Baumartz, Alexander Mehler, Mevlüt Bagci, Maxim Konca

原文摘要

Automated essay scorers (AES) based on large language models (LLMs) promise human-level accuracy. However, training, fine-tuning, and evaluating LLMs for specific assessment tasks is costly. In particular, the question arises whether these AES should be trained for specific domains or whether models based on generic LLMs might be sufficient. Human annotations, by contrast, are often more valid but far more expensive. This study evaluates whether ensembles of AES systems based on general-purpose LLMs can approximate human ratings of medical-education writing tasks by leveraging encoded linguistic and conceptual knowledge. The goal is to assess whether general-purpose models are applicable for assisting or potentially replacing human annotation in this domain. We experiment with 11 AES systems and 8 different LLMs and evaluate 668 student responses to 12 tasks, using \numprint{2076} evaluations by 7 human annotators. These ratings are aggregated to create consistent performance levels that can be used as target classes for classifiers trained on AES outputs. Our results show that learner classes based on human ratings can be accurately reconstructed using a linear SVM trained on a small set of AES features. A single classifier generalized well across tasks and domains, with reasoning and content features being most predictive. This suggests that out-of-the-box LLMs, even without domain-specific fine-tuning, can support automatic essay grading in medical education. While individual AES systems are insufficient alone, their combination approximates human evaluators, pointing to a reduced need for manual scoring or its supplementation with automated methods.

cs.CL

新论文 · v1arXiv:2611.21698

FACE: Flow-Guided Angular Channel Estimation Framework for mmWave MIMO Systems

Samitha Gunarathne, Eslam Eldeeb, Nurul Huda Mahmood, Matti Latva-aho

原文摘要

Acquiring accurate channel state information (CSI) is essential for millimeter-wave (mmWave) multi-input multi-output (MIMO) systems. However, it remains challenging due to high-dimensional channels, limited pilot observations, and measurement noise. To address these challenges, we propose flow-guided angular channel estimation (FACE), a novel channel estimation framework based on flow matching, an emerging generative modeling paradigm that learns continuous transport dynamics between probability distributions. FACE exploits the inherent angular-domain sparsity of mmWave channels and incorporates the learned flow matching model as a generative channel prior. Numerical results show that FACE consistently achieves higher channel estimation accuracy than both conventional estimators and diffusion-based generative estimator across diverse channel and system configurations. Moreover, FACE maintains high estimation accuracy with fewer pilot observations and achieves an average achievable rate close to the perfect-CSI benchmark. Compared with diffusion-based estimation, FACE requires substantially fewer sampling steps while achieving superior or comparable estimation accuracy, reducing inference time and computational cost. Overall, the proposed framework demonstrates the potential of flow matching for accurate, pilot-efficient, and computationally efficient mmWave MIMO channel estimation.

cs.LG

新论文 · v1arXiv:2611.21206

PQC Migration: End-to-End Performance Profiling of Post Quantum Cryptographic Algorithms in mTLS

Rahul, Manjesh K. Hanawal, Sameer G. Kulkarni, Saravanan V, K. Tirumaleswar Reddy

原文摘要

Emerging quantum computers threaten to break the classical key exchange algorithms. To safeguard against these threats, migration to new quantum-safe algorithms is underway and expected to be expedited. Though resistant to attacks, the quantum safe algorithms may impact the real-world performance due to their large key lengths. In this work, we analyze the end-to-end performance of the Transport Layer Security (TLS) protocol with the post-quantum (PQ) and hybrid cryptographic algorithms standardized by NIST. We develop a test setup to evaluate the performance of TLS connections between a client and a server under various network conditions. We focus on the impact of the new algorithms on the speed (throughput, connection time) and sustainability (energy, CPU cycles) of TLS 1.3. Our analysis demonstrates that the standardized post-quantum algorithms do not introduce any measurable performance degradations in TLS 1.3, despite their larger key lengths. However, hybrid algorithms introduce noticeable performance degradation, which becomes increasingly pronounced under adverse network conditions. Our analysis suggests that migration to hybrid algorithms should be ``network condition aware'' as poor network conditions can severely degrade performance in TLS 1.3. We provide anonymised link to the code and a web interface to analyse performance of various combination of algorithms.

cs.CR

新论文 · v1arXiv:2611.21203

Decision-Relevant Forecast Calibration for Predictive Control

Victor M. Preciado

原文摘要

Calibration evidence for a predictive controller need not remain valid when operating-context frequencies change. We develop an audit and certification procedure for a declared family of context-density shifts, with a frozen predictor/controller and unchanged conditional outcome law. Coverage requires calibrated tests to span density-weighted selected-plan sensitivities. A residual Gram matrix yields the minimum augmentation rank and an optimal test budget for uncovered $L_2$ geometry. This optimum need not minimize bias over probability-feasible conditional means. For finite contexts, an asymmetric absolute-loss dual and directional linear programs enclose the continuous worst-shift feasible bias. Intersecting simultaneous moment intervals with the conditional-mean domain gives a further certificate and a linear program for selecting an operating shift after validation. In a designed 240-context study, two spectral directions remove at least 80.8% of the feasible worst-case bias. For a separate misspecified logistic forecaster, matched sensitivity and generic corrections reduce expected cost by 2.590% and 3.239%, with the generic method winning all 200 refits. On 100,000 new episodes, both corrections certify absolute evaluation bias below the illustrative tolerance 0.04 throughout a declared shift interval, while the base certifies no shift. This outcome is highly predictable under the known synthetic law used for design. The probability-feasible certificate does not improve on the moment rectangle in this sample. The contribution is reusable evaluation and audit capability with explicit geometric and statistical limits, rather than recalibration superiority. Evidence uses synthetic finite contexts, known frequencies and committed plans; evaluation bias, oracle regret, constraint risk and feedback deployment remain distinct.

eess.SY · cs.LG

版本更新 · v2arXiv:2611.17060

Ropedia Xperience 2: A Gold Standard for Scaling Physical AI Environments and Behaviors

Hao Li, Fangzhou Hong, Zhaoxi Chen, Hui En Pang, Kairui Hu, Mingze Sun, Runmao Yao, Xinhao Ji 等 22 位作者

原文摘要

Large foundation models such as GPT-6 Astra and Claude Fable 5.1 have recently been shown to control robots directly from visual observations. They nevertheless remain brittle on long-horizon, contact-rich tasks in real-world environments. Further progress depends on two complementary resources: demonstrations of human behavior that supervise how tasks are performed, and interactive environments in which policies can be rolled out, evaluated, and improved in closed loop. Language models have advanced along well-characterized scaling laws; the corresponding question for physical AI is how capability scales with environments and behaviors, which first requires identifying which data to scale and how to produce and reuse such data continuously. To address this question, we introduce Ropedia Xperience 2, comprising 1,589 released metric-scale 3D scans and a collection of 4,230 egocentric recordings of 74 tasks (131 hours). We further show that each capture can be repurposed for robot learning: scenes are converted into physically interactive simulation environments, human motion is retargeted to whole-body robot trajectories, and additional robot behaviors are obtained through virtual teleoperation and automated long-horizon trajectory generation. We propose this joint capture of environments and behaviors as a gold standard for physical AI data collection, laying a data foundation for establishing scaling laws over environments and behaviors.

cs.CV

新论文 · v1arXiv:2611.21198

Cooperative Mobile Manipulation using Learned Dynamical Systems

Saray Bakker, Wendelin Böhmer, Javier Alonso-Mora

原文摘要

Cooperative mobile manipulation requires robots to coordinate their motions while ensuring whole-body collision avoidance and preserving task-relevant behaviors. We propose CALDS (Cooperative Adaptation of Learned Dynamical Systems), a framework that enables mobile manipulators to adapt learned single-agent behaviors for coordinated execution. Building on individual dynamical systems (DSs) learned from demonstrations, CALDS enables cooperative adaptation through two key components: Cooperative Policy Selection, which identifies compatible combinations of learned policies, and Temporal Adaptation, which ensures collision-free cooperation while preserving the spatial structure and convergence properties of the original behaviors. We evaluate CALDS across three diverse scenarios: assistive dressing, restaurant cleaning and serving, and drawer opening and object placement. Quantitative results demonstrate its effectiveness in improving task success and enabling safe, coordinated simultaneous execution while maintaining fidelity to demonstrated behaviors.

cs.RO

新论文 · v1arXiv:2611.21196

Determining Concentration from Raman Spectroscopy with Manifold Learning and Y-shaped Conformal Autoencoders

Anna Taka, Eleni D. Koronaki, Ioannis G. Kevrekidis, Eike Cramer, Joern Viell, Alexander Mitsos

原文摘要

Monitoring of chemical processes is crucial in chemical engineering, motivating sophisticated process analytical technology (PAT). In-line analytics are advantageous, with Raman spectroscopy being an established and reliable tool in various applications, from which many properties can be determined. We recently [Koronaki et al., AIChEJ 2024] proposed determining polymer particle size from Raman spectroscopic data, by combining Diffusion Maps (a manifold learning method) with a Y-shaped Conformal Autoencoder Neural Network. Herein, we extend this workflow to predict liquid concentrations, allowing also for multiple target predictions from the same model. We apply the methods on two datasets consisting of Raman measurements and concentration measurements: one from microgel polymerization and one from citric acid dissociation. The single-target concentration prediction results, where one target is determined from each model at a time, are accurate. Comparing against dedicated methods such as Indirect Hard Modeling (IHM) is not straightforward, due to the different validation strategies and the considered spectra range; nevertheless, it is fair to say that IHM performs slightly better than the proposed workflow. We further compare this approach with other regression methods, such as artificial neural networks, extreme gradient boosting and partial least squares, and the results show that the Conformal Autoencoders achieve higher predictive performance, demonstrating that the workflow can also be used for concentration determination. Finally we find that single-target concentration predictions are more accurate than the multiple target predictions for both case studies.

cs.LG

新论文 · v1arXiv:2611.21190

Linear Network Coding for Sum All-Reduce over Ring Networks

Yi Chen, Zhuoqi Tu, Shenghao Yang

原文摘要

Sum All-Reduce is a fundamental collective communication primitive in which each node holds a local vector and every node must recover the componentwise sum of all vectors. In distributed training, this operation is often implemented over a ring topology using a reduce--multicast scheme. We study sum All-Reduce on directed rings via linear network coding, which includes reduce--multicast as a special case. We give a necessary and sufficient feasibility condition in terms of the coding matrices. For a ring with $N\geq 3$ nodes and unit-capacity links, where each node holds a vector containing $K$ symbols, we prove that the number of time units $T$ required by any feasible linear code satisfies $T\ge\max\bigl\{K+N-2,\ \lceil\frac{2(N-1)}{N}K\rceil\bigr\}$, giving a rate $K/T$ of at most $N/(2(N-1))$. The bound is tight over every field. When $K\ge N$, an interleaved reduce--multicast code achieves the bound; when $K<N$ and $K$ divides $N$, a reduce--partial-multicast code does. When $K<N$ and $K$ does not divide $N$, no coordinate-wise code---in particular, no reduce--multicast code---achieves the bound, while an explicit linear network code does.

cs.IT

版本更新 · v2arXiv:2606.31411

Linguistic Bias Mitigation for Spoofing Detection via Gradient Reversal and A Variational Information Bottleneck

Anh-Tuan Dao, Driss Matrouf, Mickael Rouvier, Nicholas Evans

原文摘要

Rapid advancements in generative speech technology have compromised the reliability of voice biometrics. While current spoofing detectors excel when assessed under in-domain conditions, generalisation to out-of-domain settings is often poor. We show that this can be due to linguistic bias. A reliance on linguistic cues observed in training data can then compromise robustness to cross-data. We propose a linguistic-invariant spoofing detection framework utilizing teacher-student adversarial learning. The linguistic-aware teacher model, pre-trained on linguistic content of an external dataset, guides the student detector via gradient reversal to minimize the linguistic information. To prevent the inadvertent removal of non-linguistic cues, we incorporate a Variational Information Bottleneck to enable suppression of principal cues. Across nine DF Arena datasets, our method achieves up to a 36.2% relative reduction in the EER compare to the baseline.

cs.CL · cs.LG

版本更新 · v2arXiv:2611.03330

The SIMPLE Law for Energy Cost of LLM Inference

Hiari Pizzini Cavagna, Andrea Proia, Daniele Cesarini, Zeynep Kiziltan, Andrea Bartolini

原文摘要

Generative AI is billed per token, yet tokens are not equally expensive. Each token enters the KV-cache and raises the cost of those that follow, so energy grows non-linearly along a sequence whose tokens look identical. No analytical model states what a token costs without per-LLM fitting, and no law names the factors that set it. From closed-form expressions of LLM inference's computational and memory-access complexity, we derive the SIMPLE Inference Model to Predict LLMs Energy (SIMPLE), a four-parameter law that separates prefill from decode, and decode into KV-cache and next-token computation cost. Over traces of 13 dense LLMs from several families on NVIDIA H100 GPUs, it attains R2 = 0.969 and 11.97% MAPE, 3.1-3.5x below the SOTA, without per-LLM calibration. SIMPLE explains as well as predicts: prefill energy grows quadratically with hidden dimension, so larger LLMs suffer most from long inputs; and higher GQA KV grouping reduces the per-token energy linearly relative to standard MHA. Increasing the reasoning token budget pushes output length beyond the analytically derived energy minimum, driving super-linear energy growth; accuracy rises but energy rises faster, so Intelligence per Joule peaks at an intermediate budget and falls by up to 3.14x across a 4.0x range. Moving to agentic serving raises energy per request by 13.0x, as both prefill and KV-cache terms grow with input length; prefix caching removes most of the prefill cost (-93%), but not KV-cache cost, so even a cached agentic request costs 4.8x more.

cs.AI · cs.LG · cs.PF

新论文 · v1arXiv:2611.21184

An Optimal Online Algorithm for Single-machine Scheduling with Two Agents

Peihai Liu, Kejun Zhao, Xiwen Lu

原文摘要

This paper addresses the online two-agent scheduling problem on a single machine, where jobs arrive over time. The objective is to minimize the weighted sum of the agents' makespans, denoted as $C_{\max}^A + θC_{\max}^B$, for a given priority weight $θ\ge 1$. A significant gap has persisted between the established theoretical lower bound on the competitive ratio and the performance of the best-known online algorithms. For the parameter range $1 \leq θ\leq \tfrac{1+\sqrt{5}}{2}$, we propose novel online algorithm which achieves a competitive ratio of $1+\tfracθ{1+θ}$. This result matches the theoretical lower bound.

cs.DS · cs.DM

新论文 · v1arXiv:2611.21183

Seeing Isn't Resolving: Visual and Linguistic Disambiguation with Conversational Grounding

Edoardo Cecchinato, Shun Inadumi, Angel García Contreras, Giuseppe Riccardi, Koichiro Yoshino

原文摘要

Visual grounding benchmarks usually do not consider the contextual uncertainty of real-world scenarios. In this study, we construct a novel benchmark, the Japanese evaluation for Ambiguous visual Grounding with Dialogue (JAG-D) dataset. It consists of human-robot conversational multimodal interactions covering different types of ambiguity, arising from visual and linguistic categories. We evaluate the capabilities of different state-of-the-art Visual Language Models (VLMs) for ambiguity recognition and resolution. We compare dialogue context grounding performance for visual and linguistic ambiguity resolution. Our experiments show that VLMs can substantially improve the resolution of linguistic ambiguity, especially Japanese zero anaphora, by using dialogue context. We further conduct a comparative analysis to understand whether grounding improvements arise from the semantic content of the dialogue or from its conversational structure.

cs.CL

新论文 · v1arXiv:2611.21178

TRACK: Tool-Augmented Reasoning over Time Series for Fault Diagnosis in Power Management System

Erkan Bayram, Debanjan Ghosh, Li Zhu, Sambarta Dasgupta

原文摘要

Fault diagnosis in power management systems running multiple loads relies on experts manually inspecting current and voltage waveforms, which fails to scale across deployed units. We present TRACK, which pairs a time series encoder with a Small Language Model to produce structured, physics-grounded diagnostic reports, calling tools for exact, verifiable calculations. It operates on-device, keeping proprietary sensor data local and avoiding expensive API calls. We introduce a rubric-based framework that evaluates these reports by scoring structure, signal fidelity, physics reasoning, diagnosis, and actionability through per-criterion verdicts. On a 15-class multi-load simulated benchmark, TRACK achieves over 95% fault-category accuracy and generalizes to out-of-distribution faults, whereas the best encoder-probing baseline reaches only 58% accuracy.

cs.LG · cs.AI · cs.CL · eess.SY

版本更新 · v2arXiv:2611.17007

Skill Adapter: Personalized Modification and Online Routing of General-Purpose Skills

Qijia Zhuang, Yanyin Gou, Wenbin Ma, Yuxiang Ren

原文摘要

Large language model (LLM) agents use external skills to reuse task procedures and experience with tools. General-purpose skills describe shared execution procedures, while specific tool choices, operating practices, and output requirements vary across users. In long-horizon tasks, subgoals, available inputs, and environmental states change as execution progresses. Agents must select currently applicable skills from an expanding skill pool and prepare candidates for subsequent steps. We propose Skill Adapter, which enables personalized modification and online routing of general-purpose skills. The system distills local adjustments from user requirements and execution feedback and stores them as Adapters linked to general-purpose skills. The routing agent selects general-purpose skills and applicable Adapters based on the current subtask, execution progress, and environmental feedback, incrementally maintaining a task-level online skill library that the main LLM reads on demand. The system preloads candidates for subsequent nodes and rechecks their applicability once actual inputs are determined. Reinforcement learning optimizes the two types of decisions using modification gains and routing outcomes, respectively. We evaluate Skill Adapter on ALFWorld, WebShop, and SkillsBench. Compared with SkillRL, it improves task success rates by an average of 11.43 percentage points and reduces mean execution time on all three benchmarks. Skill Adapter retains its performance advantage when the skill pool grows to 1,000 skills.

cs.AI

新论文 · v1arXiv:2611.21177

RAGE: Residual-Adaptive Gating for Flow-Based Image Editing

Ruofeng Mei, Xiangyu Xu

原文摘要

Recent inversion-free, flow-based image editing methods evolve the source image toward the target via a differential velocity field. We observe that this velocity field does not explicitly account for which regions should be preserved, leading to unintended drift in non-editing regions. To address this, we formulate source fidelity control as a stochastic optimal control problem in which a selective fidelity term on the editing residual is enforced along the editing trajectory. Solving this problem and approximating its solution yields the Residual Gating Principle: each latent component is pulled back toward the source with a gain that decays exponentially with the local residual magnitude, so that small deviations are suppressed while large, semantically meaningful edits pass through unimpeded. The proposed method, RAGE, requires only one lightweight gated interpolation per sampling step, involves no mask estimation, attention maps, retraining, or architecture-specific modifications, and is applied on top of FlowEdit-based editors. On PIE-Bench, RAGE achieves the best overall rank among 17 methods and substantially improves background preservation on both Stable Diffusion 3 and FLUX backbones while maintaining competitive editing quality.

cs.CV · cs.LG

新论文 · v1arXiv:2611.21176

Scientific Coding with Large Language Models: Evidence from Finite Element Analysis

Erfan Hamdi, Joel Shor, Emma Lejeune

原文摘要

Large language models (LLMs) offer a potential route to automating scientific programming, but their usefulness depends on consistently producing numerically correct code. We extend FEM-Bench to 17 finite element tasks involving nonlinear and history-dependent mechanics and structural dynamics, and evaluate 15 models on implementation and test generation without execution feedback. The best model succeeds in 87.1% of single-pass implementation trials, but rankings differ between tracks. The leading implementer achieves 49.9% joint test success, while the leading tester reaches 82.6%. For six models, we compare single-pass generation with two agentic harnesses that allow code execution and revision. Agentic execution solves some tasks that never succeeded in single-pass trials and improves consistency on others, but gains vary across models and harnesses, and regressions also occur. Claude Opus 5 with SimpleAgent reaches 96.5% Pass@3, the probability of at least one success in three attempts, but 77.1% Pass$^3$, the probability that all three succeed. No model-harness combination exceeds 78.8% Pass$^3$. To understand why agents submit incorrect code after passing their own checks, we analyze 249 failed agentic trajectories and identify six categories of self-verification failure. The most common are inadequate test coverage (73.1%) and reliance on generic consistency checks (70.3%), and in 28.5% of cases the tests share an error with the implementation. Motivated by these findings, an exploratory configuration with separate coder and tester agents achieves the highest observed scores, with 97.6% Pass@3 and 88.8% Pass$^3$. These results identify consistency across attempts and independent verification as the main obstacles to trusting LLM-generated scientific code.

cs.SE · cs.AI · cs.CE

新论文 · v1arXiv:2611.21175

A COMPASS to the Galaxy: Joint nucleosynthetic yield set comparison and parameter inference in Galactic chemical evolution

Berkay Günes, Giuseppe Viterbo, Sven Buder, Tobias Buck

原文摘要

Galactic Chemical Evolution (GCE) models connect stellar nucleosynthesis to observed abundance patterns, but their interpretation requires simultaneous parameter inference and comparison between competing stellar yield prescriptions. This remains challenging due to intractable likelihoods and growing dataset complexity. Using compass, a novel simulation-based inference framework, we perform, for the first time, amortized joint Bayesian model comparison and parameter inference for GCE models with intractable likelihoods within a unified architecture. Compass employs a conditional score-based diffusion model with a transformer architecture to learn both the posterior and likelihood directly from forward simulations, enabling efficient inference and likelihood-based ranking of competing nucleosynthetic yield sets. In controlled tests, compass identifies the generating yield set and recovers ground-truth parameters with accuracy comparable to existing simulation-based inference (SBI) methods, while reducing computational cost by orders of magnitude. Applied to Milky Way data, the framework favors a hybrid NuGrid + IllustrisTNG yield prescription, confirmed by an independent classifier, and infers an initial mass function slope of $α_{\rm IMF} = -2.65 \pm 0.02$ and a SN$_{\rm Ia}$ normalization of $\log(N_{\rm Ia}) = -2.81 \pm 0.02$, with a systematic spread of $\approx 0.3$ in $α_{\rm IMF}$ between the best-ranked yield sets, consistent with independent constraints. The attention mechanism confirms that the inferred parameters are driven by physically meaningful abundance signatures. By unifying amortized parameter inference, Bayesian model comparison, and interpretability in a single probabilistic framework, compass provides a scalable approach for precision chemical evolution studies with current and forthcoming spectroscopic surveys.

astro-ph.GA · astro-ph.IM · cs.LG · physics.data-an

版本更新 · v2arXiv:2606.08678

Speaker-Invariant Representation Learning for Spoofing Detection via Gradient Reversal and A Variational Information Bottleneck

Anh-Tuan Dao, Driss Matrouf, Mickael Rouvier, Nicholas Evans

原文摘要

Sophisticated generative speech technology can undermined the reliability of voice biometrics. While spoofing detection systems excel when assessed under in-domain conditions, generalisation to out-of-domain settings is often poor. In this paper, we show that such issues could be caused by speaker bias, where models learn individual voice traits rather than markers of manipulation or generation. We propose a teacher-student framework for speaker-invariant spoofing detection that disentangles identity without requiring speaker labels. We leverage a pre-trained speaker recognition teacher to guide a student model via a gradient reversal layer. To control the balance between suppressing cues related to voice identity with the preservation of those related to spoofing detection, we integrate a Variational Information Bottleneck. Evaluations across nine datasets show our model achieves a 25.7% relative reduction to the EER compared to the MHFA baseline.

cs.SD · cs.LG

版本更新 · v2arXiv:2610.11161

VGGT-VLA-World: Intent-Conditioned 3D World Evolution for Autonomous Driving

Zhaoyang Liu, Zhigui Lin, Kun Jiang, Tianhui Li, Changye Liu, Ziying Song, Diange Yang

原文摘要

VGGT provides a strong foundation for geometry-centric world models by recovering unified 3D scene geometry from visual observations. Although recent extensions enable temporal 3D prediction, their future evolution remains weakly conditioned on driving intentions and actions, limiting their ability to model alternative action-dependent futures. We propose \textbf{VGGTWorld-VLA}, an intention-conditioned extension of VGGT-World for controllable 3D world evolution in autonomous driving. First, we introduce an action--semantic conditioning mechanism that injects complementary driving semantics and ego-motion representations into the future-token stream, enabling different future geometry predictions for the same observed scene under alternative ego actions. Second, we develop a geometry--language--action bridge that adapts historical geometry, VLA semantic features, and maneuver and trajectory representations for joint conditioning of future geometry prediction. We evaluate future geometry prediction on NAVSIM, with ablation studies examining the contributions of semantic and action conditioning. Compared with baseline, our method reduces future-depth RMSE by 13.7\% and improves $δ_1$ by 4.11 percentage points on NAVSIM motion subset. Controlled action-input interventions further show that the model uses the supplied action information. These results support semantic and action conditioning for controllable VGGT-based world prediction in autonomous driving.

cs.CV · cs.RO

新论文 · v1arXiv:2611.21165

A Matter of Stages: Energy-Latency Trade-offs in Multimodal LLM Serving

Yifan Sun, Jiantong Jiang, Zhexiang Zhang, Xu Bai, Feng Liu, Rajkumar Buyya, Adel N. Toosi

原文摘要

Online serving of multimodal large language models requires deciding whether to separate visual encoding (E), prefill (P), and decode (D) across GPUs and how to configure their frequencies. EPD disaggregation enables independent stage control, but it also changes communication and queueing behavior. Lowering one stage's frequency can increase queueing and prolong execution across the EPD pipeline. Consequently, reducing power at one stage does not necessarily reduce total GPU energy per request. In this paper, we systematically characterize fused and EPD-disaggregated serving across four multimodal models under a fixed three-GPU budget, followed by single-stage and uniform downclocking experiments across 18 workload combinations. We find that disaggregation reduces GPU energy per successful request by 13.3-37.0% at nearly identical output-token throughput, but increases mean TTFT. Downclocking exhibits distinct stage sensitivities, and aggressive frequency reductions can offset power savings through queueing and longer execution. Across the evaluated workloads, uniform downclocking to 80% reduces GPU energy by 11.7-18.6% with modest mean-latency changes. Guided by these findings, we develop a budgeted frequency exploration strategy to select energy-efficient configurations under latency constraints. Evaluation shows that four probes can identify the lowest-energy feasible configuration among 25 measured candidates across all 288 evaluated positive-margin SLO scenarios. These results provide practical guidance for energy-efficient multimodal serving.

cs.DC

版本更新 · v6arXiv:2605.18267

SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation

Longtao Jiang, Jianmin Bao, Zhendong Wang, Xin Tao, Zhihui Li, Xiaojun Chang

原文摘要

Normalizing flows (NFs) provide exact likelihoods and deterministic invertible sampling, but have historically lagged behind diffusion models for large-scale image generation. We identify a key obstacle: NFs are required to learn a single invertible transport over the full ambient space, making them highly sensitive to high-dimensional representations. This leads to a semantic-capacity mismatch in modern visual representation spaces, where semantic information is compact but encoded in overcomplete features. We propose SRC-Flow, which introduces a Semantic Representation Compressor (SRC) to compact high-dimensional RAE features into a low-dimensional semantic space before flow modeling and preserve reconstruction through the frozen RAE decoder. This compact space reduces the modeling burden of NFs and enables effective likelihood-based generation in semantic representation space. We further adopt constant noise regularization tailored to the fixed unconditional bijection learned by flows. On ImageNet 256x256 and 512x512, SRC-Flow achieves state-of-the-art generation quality among normalizing flow methods, with gFID scores of 1.65 and 2.07 under classifier-free guidance, while retaining exact likelihood computation in the compact semantic representation space and deterministic invertible sampling at the flow level. Codes and models will be released in the future.

cs.CV