arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

期刊&会议

AAAI Conference on Artificial Intelligence · 会议 · Artificial Intelligence

2026-03-23 至 2026-03-23 共收录 17
2603.19831 2026-03-23 eess.AS cs.AI cs.MM

Gesture2Speech: How Far Can Hand Movements Shape Expressive Speech?

Gesture2Speech: 手部动作能多大程度上塑造表现性语音?

Lokesh Kumar, Nirmesh Shah, Ashishkumar P. Gudmalwar, Pankaj Wasnik

AI总结 本文提出Gesture2Speech框架,利用视觉手势线索调节合成语音的语调,通过多模态MoE架构动态融合语言内容和手势特征,提升语音自然度和手势与语调的同步性。

Comments Accepted at The 2nd International Workshop on Bodily Expressed Emotion Understanding (BEEU) at AAAI 2026 [non-archival]

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.19579 2026-03-23 cs.AI cs.LG

PA2D-MORL: Pareto Ascent Directional Decomposition based Multi-Objective Reinforcement Learning

PA2D-MORL:基于帕累托上升方向分解的多目标强化学习

Tianmeng Hu, Biao Luo

AI总结 本文提出PA2D-MORL方法,通过帕累托上升方向选择标量化权重并计算多目标策略梯度,实现高效多目标问题分解与策略优化,提升帕累托策略集的逼近质量。

Comments AAAI 2024

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 38(11), 12547-12555, 2024

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.24903 2026-03-23 cs.CV cs.CE

FinMMDocR: Benchmarking Financial Multimodal Reasoning with Scenario Awareness, Document Understanding, and Multi-Step Computation

FinMMDocR:基于场景意识、文档理解与多步骤计算的金融多模态推理基准测试

Zichen Tang, Haihong E, Rongjin Li, Jiacheng Liu, Linwei Jia, Zhuodi Hao, Zhongjun Yang, Yuanze Li, Haolin Tian, Xinyi Hu, Peizhi Zhao, Yuan Liu, Zhengyu Wang, Xianghe Wang, Yiling Huang, Xueyuan Lin, Ruofei Bai, Zijian Xie, Qian Huang, Ruining Cao, Haocheng Gao

AI总结 FinMMDocR通过引入场景意识、文档理解和多步骤计算,推动多模态大语言模型在现实金融场景中的推理能力提升,其包含12种隐含金融场景、9类丰富文档及平均11步推理任务。

Comments Accepted by AAAI-26 Main Track

Journal ref Proc. AAAI 2026 (Vol. 40, No. 30), pages 25858-25866

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.17910 2026-03-23 cs.CL

L2V-CoT: Cross-Modal Transfer of Chain-of-Thought Reasoning via Latent Intervention

L2V-CoT:通过潜在干预实现链式推理的跨模态转移

Yuliang Zhan, Xinyu Tang, Han Wan, Jian Li, Ji-Rong Wen, Hao Sun

AI总结 本文提出L2V-CoT方法,通过潜在干预将链式推理从LLM转移到VLM,利用低频潜在表示提升多步推理能力,实验表明优于无训练基线和监督方法。

Comments AAAI 2026 oral

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.08916 2026-03-23 cs.CL

HalluClean: A Unified Framework to Combat Hallucinations in LLMs

HalluClean:一种用于对抗大语言模型幻觉的统一框架

Yaxin Zhao, Yu Zhang

AI总结 HalluClean通过规划、执行和修订三阶段推理增强方法,有效检测并修正LLM生成文本中的幻觉,提升事实一致性并优于基线方法。

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 40(42), 36092-36100 (2026)

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.08453 2026-03-23 cs.SI

Whose Values? Measuring the (Subjective) Expression of Basic Human Values in Social Media

谁的价值?在社交媒体中测量(主观)基本人类价值观的表达

Ziv Epstein, Farnaz Jahanbakhsh, Tiziano Piccardi, Isabel Gallegos, Dora Zhao, Johan Ugander, Michael Bernstein

AI总结 本文基于Schwartz价值观体系,提出一种规模化测量社交媒体中价值观表达的框架,通过个性化校准注释提升预测准确性,揭示人类价值观测量的新方法。

Comments Proceedings of the International AAAI Conference on Web and Social Media. 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.07798 2026-03-23 cs.CV

Divide-and-Conquer Decoupled Network for Cross-Domain Few-Shot Segmentation

分治解耦网络用于跨域少样本分割

Runmin Cong, Anpeng Wang, Bin Wan, Cong Zhang, Xiaofei Zhou, Wei Zhang

AI总结 本文提出分治解耦网络,通过对抗学习和对比学习解耦特征,提升跨域少样本分割的泛化能力与适应速度。

Journal ref AAAI 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2510.14184 2026-03-23 cs.LG cs.AI cs.CL

MAFA: A Multi-Agent Framework for Enterprise-Scale Annotation with Configurable Task Adaptation

MAFA:一种用于企业级标注的多智能体框架,支持可配置的任务适应

Mahmood Hegazy, Aaron Rodrigues, Azzam Naeem

AI总结 MAFA通过可配置的多智能体协作解决金融行业标注积压问题,实现86%的人工标注一致性,每年节省5000小时人工标注工作,提升标注准确率和效率。

Journal ref AAAI Conference on Artificial Intelligence 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2510.05138 2026-03-23 cs.CL

LiRA: A Multi-Agent Framework for Reliable and Readable Literature Review Generation

LiRA:一种用于生成可靠且易读文献综述的多智能体框架

Gregory Hok Tjoan Go, Khang Ly, Anders Søgaard, Amin Tabatabaei, Maarten de Rijke, Xinyi Chen

AI总结 LiRA通过多智能体协作流程生成综合且全面的文献综述,优于现有基线方法,在写作和引用质量上表现更优,同时保持与人工写作相似度。

Comments Published at the 40th AAAI Conference on Artificial Intelligence. Please cite the published version here: https://ojs.aaai.org/index.php/AAAI/article/view/41489

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.19361 2026-03-23 cs.AI cs.CV cs.LG cs.LO

Consistency-based Abductive Reasoning over Perceptual Errors of Multiple Pre-trained Models in Novel Environments

基于一致性的多预训练模型感知误差的归纳推理

Mario Leiva, Noel Ngu, Joshua Shay Kricheli, Aditya Taparia, Ransalu Senanayake, Paulo Shakarian, Nathaniel Bastian, John Corcoran, Gerardo Simari

AI总结 本文提出利用多个预训练模型减少感知误差的归纳推理方法,通过逻辑程序编码预测和错误检测规则,采用整数规划和启发式搜索算法提升预测覆盖度和一致性,实验表明在复杂分布偏移场景中性能优于单一模型和传统集成基线。

Comments Accepted to AAAI 2026. Code available at https://github.com/lab-v2/EDCR_PyReason_AirSim

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.03424 2026-03-23 cs.LG cs.AI

Framework GNN-AID: Graph Neural Network Analysis Interpretation and Defense

图神经网络分析解释与防御框架 GNN-AID

Kirill Lukyanov, Mikhail Drobyshevskiy, Georgii Sazonov, Mikhail Soloviov, Ilya Makarov

机构 * ISP RAS Research Center for Trusted Artificial Intelligence(俄罗斯科学院可信人工智能研究信息与处理中心) Ivannikov Institute for System Programming of the Russian Academy of Sciences(俄罗斯科学院伊万诺夫系统编程研究所) Moscow Institute of Physics and Technology (National Research University)(莫斯科物理技术学院(国家研究大学)) Lomonosov Moscow State University(罗蒙诺索夫莫斯科国立大学) AIRI

AI总结 本文提出GNN-AID框架,旨在解决图数据中可解释性与鲁棒性结合的问题,提供分析、解释和防御工具,支持多种攻击与防御方法,并提供可视化和MLOps技术以提升可重复性。

Journal ref 2026 Proceedings of the AAAI Conference on Artificial Intelligence, 40(48), 41634-41636

详情

展开后加载摘要…

URL PDF HTML 收藏
2502.15851 2026-03-23 cs.CL cs.AI

Control Illusion: The Failure of Instruction Hierarchies in Large Language Models

控制幻觉:大型语言模型中指令层级的失效

Yilin Geng, Haonan Li, Honglin Mu, Xudong Han, Timothy Baldwin, Omri Abend, Eduard Hovy, Lea Frermann

AI总结 研究探讨了大型语言模型中指令层级机制的有效性,发现模型在处理简单格式冲突时难以保持一致的优先级,且系统/用户提示分离方法无法建立可靠层级,社会层级框架对模型行为影响更大。

Comments Accepted to AAAI-26 Main Technical Track Proceedings

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 40(36): 30816-30824, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.16546 2026-03-23 cs.CL cs.AI

DanceHA: A Multi-Agent Framework for Document-Level Aspect-Based Sentiment Analysis

DanceHA:一种用于文档级基于方面的情感分析的多智能体框架

Lei Wang, Min Huang, Eduard Dragut

AI总结 本文提出DanceHA多智能体框架,用于文档级基于方面的情感强度分析,通过分解任务和人机协作提升处理复杂任务的能力,并展示了其在多领域数据集上的有效性。

Journal ref AAAI 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.09792 2026-03-23 cs.LG cs.MA

Beyond Monotonicity: Revisiting Factorization Principles in Multi-Agent Q-Learning

超越单调性:重新审视多智能体Q学习中的分解原理

Tianmeng Hu, Yongzheng Cui, Rui Tang, Biao Luo, Ke Li

机构 * Department of Computer Science, University of Exeter, U.K.(埃克塞特大学计算机科学系) School of Automation, Central South University, China(中南大学自动化学院)

AI总结 本文通过动力系统分析非单调值分解,证明非单调分解能可靠恢复IGM最优解并优于单调基线。

Comments Accepted at AAAI 2026

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 40(26), 21876-21884, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.08015 2026-03-23 cs.CV cs.AI

Invisible Triggers, Visible Threats! Road-Style Adversarial Creation Attack for Visual 3D Detection in Autonomous Driving

不可见的触发器,可见的威胁!面向自动驾驶视觉3D检测的路式对抗生成攻击

Jian Wang, Lijun He, Yixing Yong, Haixia Bi, Fan Li

AI总结 本文提出AdvRoad,通过生成自然道路风格的对抗性贴纸,提升自动驾驶系统在视觉3D检测中的鲁棒性,实验表明其在不同检测器和场景中均有效。

Comments Accepted by the AAAI 2026 (Main Track)

Journal ref AAAI Conference on Artificial Intelligence, 40(12), 9903-9911. (2026)

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.07889 2026-03-23 cs.CV cs.AI

Generating Sketches in a Hierarchical Auto-Regressive Process for Flexible Sketch Drawing Manipulation at Stroke-Level

通过分层自回归过程生成草图以实现灵活的笔触级草图操作

Sicong Zang, Shuhui Gao, Zhijun Fang

机构 * Sicong Zang, Shuhui Gao, Zhijun Fang

AI总结 本文提出分层自回归生成过程,实现灵活的笔触级草图操作,通过自回归方式生成笔触嵌入并引导模型生成合适笔触。

Comments Accepted by AAAI 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2508.18839 2026-03-23 cs.LG cs.CR

DRMD: Deep Reinforcement Learning for Malware Detection under Concept Drift

DRMD:基于概念漂移的深度强化学习恶意软件检测

Shae McFadden, Myles Foley, Mario D'Onghia, Chris Hicks, Vasilios Mavroudis, Nicola Paoletti, Fabio Pierazzi

机构 * King’s College London(伦敦国王学院) The Alan Turing Institute(艾伦·图灵研究所) University College London(伦敦大学学院)

AI总结 本文提出DRMD,通过深度强化学习在Android恶意软件检测中应对概念漂移,提升检测性能和抗漂移能力,实验显示其在时间感知评估中优于传统方法。

Comments The Fortieth AAAI Conference on Artificial Intelligence (AAAI-26)

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 40, No. 2, pp. 854-862, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏