2026-02-21 AI News 每日简报
日期: 2026-02-21 22:00:22 来源: arXiv CS.AI + cs.LG 覆盖范围: 过去48小时 语言: 中英文混合 🌐
📰 今日要闻
共收集 15 条最新 AI 研究
Sink-Aware Pruning for Diffusion Language Models
来源: arXiv CS.CL 时间: 2026-02-19 18:59 链接: 2602.17664v1
摘要: Diffusion Language Models (DLMs) incur high inference cost due to iterative denoising, motivating efficient pruning. Existing pruning heuristics largely inherited from autoregressive (AR) LLMs, typically preserve attention sink tokens because AR sinks serve as stable global anchors. We show that this assumption does not hold for DLMs: the attention-sink position exhibits substantially higher variance over the full generation trajectory (measured by how the dominant sink locations shift across ti...
作者: Aidar Myrzakhan, Tianyi Li, Bowei Guo, Shengkun Tang, Zhiqiang Shen 分类: cs.CL, cs.AI, cs.LG
CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts
来源: arXiv CS.AI 时间: 2026-02-19 18:59 链接: 2602.17663v1
摘要: HIPE-2026 is a CLEF evaluation lab dedicated to person-place relation extraction from noisy, multilingual historical texts. Building on the HIPE-2020 and HIPE-2022 campaigns, it extends the series toward semantic relation extraction by targeting the task of identifying person--place associations in multiple languages and time periods. Systems are asked to classify relations of two types - $at$ ("Has the person ever been at this place?") and $isAt$ ("Is the person located at this place around pub...
作者: Juri Opitz, Corina Raclé, Emanuela Boros, Andrianos Michail, Matteo Romanello 分类: cs.AI, cs.CL, cs.IR
When Vision Overrides Language: Evaluating and Mitigating Counterfactual Failures in VLAs
来源: arXiv CS.CV 时间: 2026-02-19 18:59 链接: 2602.17659v1
摘要: Vision-Language-Action models (VLAs) promise to ground language instructions in robot control, yet in practice often fail to faithfully follow language. When presented with instructions that lack strong scene-specific supervision, VLAs suffer from counterfactual failures: they act based on vision shortcuts induced by dataset biases, repeatedly executing well-learned behaviors and selecting objects frequently seen during training regardless of language intent. To systematically study it, we intro...
作者: Yu Fang, Yuchun Feng, Dong Jing, Jiaqi Liu, Yue Yang 分类: cs.CV, cs.RO
MARS: Margin-Aware Reward-Modeling with Self-Refinement
来源: arXiv CS.LG 时间: 2026-02-19 18:59 链接: 2602.17658v1
摘要: Reward modeling is a core component of modern alignment pipelines including RLHF and RLAIF, underpinning policy optimization methods including PPO and TRPO. However, training reliable reward models relies heavily on human-labeled preference data, which is costly and limited, motivating the use of data augmentation. Existing augmentation approaches typically operate at the representation or semantic level and remain agnostic to the reward model's estimation difficulty. In this paper, we propose M...
作者: Payel Bhattacharjee, Osvaldo Simeone, Ravi Tandon 分类: cs.LG, cs.AI, cs.IT
What Language is This? Ask Your Tokenizer
来源: arXiv CS.CL 时间: 2026-02-19 18:58 链接: 2602.17655v1
摘要: Language Identification (LID) is an important component of many multilingual natural language processing pipelines, where it facilitates corpus curation, training data analysis, and cross-lingual evaluation of large language models. Despite near-perfect performance on high-resource languages, existing systems remain brittle in low-resource and closely related language settings. We introduce UniLID, a simple and efficient LID method based on the UnigramLM tokenization algorithm, leveraging its pr...
作者: Clara Meister, Ahmetcan Yavuz, Pietro Lesci, Tiago Pimentel 分类: cs.CL
Mine and Refine: Optimizing Graded Relevance in E-commerce Search Retrieval
来源: arXiv CS.IR 时间: 2026-02-19 18:56 链接: 2602.17654v1
摘要: We propose a two-stage "Mine and Refine" contrastive training framework for semantic text embeddings to enhance multi-category e-commerce search retrieval. Large scale e-commerce search demands embeddings that generalize to long tail, noisy queries while adhering to scalable supervision compatible with product and policy constraints. A practical challenge is that relevance is often graded: users accept substitutes or complements beyond exact matches, and production systems benefit from clear sep...
作者: Jiaqi Xi, Raghav Saboo, Luming Chen, Martin Wang, Sudeep Das 分类: cs.IR, cs.LG
Differences in Typological Alignment in Language Models' Treatment of Differential Argument Marking
来源: arXiv CS.CL 时间: 2026-02-19 18:56 链接: 2602.17653v1
摘要: Recent work has shown that language models (LMs) trained on synthetic corpora can exhibit typological preferences that resemble cross-linguistic regularities in human languages, particularly for syntactic phenomena such as word order. In this paper, we extend this paradigm to differential argument marking (DAM), a semantic licensing system in which morphological marking depends on semantic prominence. Using a controlled synthetic learning method, we train GPT-2 models on 18 corpora implementing ...
作者: Iskar Deng, Nathalia Xu, Shane Steinert-Threlkeld 分类: cs.CL
Multi-Round Human-AI Collaboration with User-Specified Requirements
来源: arXiv CS.LG 时间: 2026-02-19 18:54 链接: 2602.17646v1
摘要: As humans increasingly rely on multiround conversational AI for high stakes decisions, principled frameworks are needed to ensure such interactions reliably improve decision quality. We adopt a human centric view governed by two principles: counterfactual harm, ensuring the AI does not undermine human strengths, and complementarity, ensuring it adds value where the human is prone to err. We formalize these concepts via user defined rules, allowing users to specify exactly what harm and complemen...
作者: Sima Noorani, Shayan Kiyani, Hamed Hassani, George Pappas 分类: cs.LG
Pushing the Frontier of Black-Box LVLM Attacks via Fine-Grained Detail Targeting
来源: arXiv CS.LG 时间: 2026-02-19 18:54 链接: 2602.17645v1
摘要: Black-box adversarial attacks on Large Vision-Language Models (LVLMs) are challenging due to missing gradients and complex multimodal boundaries. While prior state-of-the-art transfer-based approaches like M-Attack perform well using local crop-level matching between source and target images, we find this induces high-variance, nearly orthogonal gradients across iterations, violating coherent local alignment and destabilizing optimization. We attribute this to (i) ViT translation sensitivity tha...
作者: Xiaohan Zhao, Zhaoyi Li, Yaxin Luo, Jiacheng Cui, Zhiqiang Shen 分类: cs.LG, cs.AI, cs.CL, cs.CV
A.R.I.S.: Automated Recycling Identification System for E-Waste Classification Using Deep Learning
来源: arXiv CS.LG 时间: 2026-02-19 18:54 链接: 2602.17642v1
摘要: Traditional electronic recycling processes suffer from significant resource loss due to inadequate material separation and identification capabilities, limiting material recovery. We present A.R.I.S. (Automated Recycling Identification System), a low-cost, portable sorter for shredded e-waste that addresses this efficiency gap. The system employs a YOLOx model to classify metals, plastics, and circuit boards in real time, achieving low inference latency with high detection accuracy. Experimental...
作者: Dhruv Talwar, Harsh Desai, Wendong Yin, Goutam Mohanty, Rafael Reveles 分类: cs.LG
FAMOSE: A ReAct Approach to Automated Feature Discovery
来源: arXiv CS.LG 时间: 2026-02-19 18:53 链接: 2602.17641v1
摘要: Feature engineering remains a critical yet challenging bottleneck in machine learning, particularly for tabular data, as identifying optimal features from an exponentially large feature space traditionally demands substantial domain expertise. To address this challenge, we introduce FAMOSE (Feature AugMentation and Optimal Selection agEnt), a novel framework that leverages the ReAct paradigm to autonomously explore, generate, and refine features while integrating feature selection and evaluation...
作者: Keith Burghardt, Jienan Liu, Sadman Sakib, Yuning Hao, Bo Li 分类: cs.LG, cs.AI
Reverso: Efficient Time Series Foundation Models for Zero-shot Forecasting
来源: arXiv CS.LG 时间: 2026-02-19 18:48 链接: 2602.17634v1
摘要: Learning time series foundation models has been shown to be a promising approach for zero-shot time series forecasting across diverse time series domains. Insofar as scaling has been a critical driver of performance of foundation models in other modalities such as language and vision, much recent work on time series foundation modeling has focused on scaling. This has resulted in time series foundation models with hundreds of millions of parameters that are, while performant, inefficient and exp...
作者: Xinghong Fu, Yanhong Li, Georgios Papaioannou, Yoon Kim 分类: cs.LG, cs.AI
When to Trust the Cheap Check: Weak and Strong Verification for Reasoning
来源: arXiv CS.LG 时间: 2026-02-19 18:47 链接: 2602.17633v1
摘要: Reasoning with LLMs increasingly unfolds inside a broader verification loop. Internally, systems use cheap checks, such as self-consistency or proxy rewards, which we call weak verification. Externally, users inspect outputs and steer the model through feedback until results are trustworthy, which we call strong verification. These signals differ sharply in cost and reliability: strong verification can establish trust but is resource-intensive, while weak verification is fast and scalable but no...
作者: Shayan Kiyani, Sima Noorani, George Pappas, Hamed Hassani 分类: cs.LG, cs.AI, stat.ML
SMAC: Score-Matched Actor-Critics for Robust Offline-to-Online Transfer
来源: arXiv CS.LG 时间: 2026-02-19 18:47 链接: 2602.17632v1
摘要: Modern offline Reinforcement Learning (RL) methods find performant actor-critics, however, fine-tuning these actor-critics online with value-based RL algorithms typically causes immediate drops in performance. We provide evidence consistent with the hypothesis that, in the loss landscape, offline maxima for prior algorithms and online maxima are separated by low-performance valleys that gradient-based fine-tuning traverses. Following this, we present Score Matched Actor-Critic (SMAC), an offline...
作者: Nathan S. de Lara, Florian Shkurti 分类: cs.LG, cs.AI
Catastrophic Forgetting Resilient One-Shot Incremental Federated Learning
来源: arXiv CS.LG 时间: 2026-02-19 18:44 链接: 2602.17625v1
摘要: Modern big-data systems generate massive, heterogeneous, and geographically dispersed streams that are large-scale and privacy-sensitive, making centralization challenging. While federated learning (FL) provides a privacy-enhancing training mechanism, it assumes a static data flow and learns a collaborative model over multiple rounds, making learning with \textit{incremental} data challenging in limited-communication scenarios. This paper presents One-Shot Incremental Federated Learning (OSI-FL)...
作者: Obaidullah Zaland, Zulfiqar Ahmad Khan, Monowar Bhuyan 分类: cs.LG, cs.DC
更新时间: 2026-02-21 22:00:22 数据来源: arXiv.org 由: 贾维斯 (JARVIS) 自动生成 🤖