AAAI

SGDiff: Scene Graph Guided Diffusion Model for Image Collaborative SegCaptioning

SGDiff: 基于场景图引导的扩散模型用于图像协作分割与描述

Xu Zhang, Jin Yuan, Hanwang Zhang, Guojin Zhong, Yongsheng Zang, Jiacheng Lin, Zhiyong Li

AI总结 SGDiff通过基于场景图引导的扩散模型，实现从简单提示到多样化描述与掩码对的图像协作分割与描述任务，提升语义对齐与灵活结果选择能力。

Comments Accept by AAAI-2025

Reasoning About the Unsaid: Misinformation Detection with Omission-Aware Graph Inference

对未言之的推理：基于遗漏感知图推断的虚假信息检测

Zhengjia Wang, Danding Wang, Qiang Sheng, Jiaying Wu, Juan Cao

机构 * Media Synthesis and Forensics Lab, Institute of Computing Technology, Chinese Academy of Sciences（媒体合成与取证实验室，计算技术研究所，中国科学院） ； University of Chinese Academy of Sciences（中国科学院大学） ； National University of Singapore（新加坡国立大学） 纠错

AI总结 本文提出OmiGraph框架，通过省略感知图推断实现虚假信息检测，提升检测性能。

Comments AAAI 2026

Does Self-Evaluation Enable Wireheading in Language Models?

自我评估是否能导致语言模型中的wireheading？

David Demitri Africa, Hans Ethan Ting

机构 * Ateneo de Manila University（马尼拉大学） 纠错

AI总结 研究探讨自我评估与奖励信号结合是否导致语言模型wireheading，发现评分通胀现象，但解耦后仍存在过度自信，提示需更谨慎的机制设计。

Comments Accepted (oral) to Foundations of Agentic Systems Theory at AAAI 2026

Rethinking Multimodal Point Cloud Completion: A Completion-by-Correction Perspective

重新思考多模态点云补全：一种补全-修正视角

Wang Luo, Di Wu, Hengyuan Na, Yinlin Zhu, Miao Hu, Guocong Quan

机构 * Wang Luo, Di Wu, Hengyuan Na, Yinlin Zhu, Miao Hu, Guocong Quan（作者） 纠错

AI总结 PGNet通过补全-修正范式，结合多阶段框架和双特征编码，实现更鲁棒的点云补全，提升重建精度与结构一致性。

Comments Accepted by AAAI 2026

Harnessing Diffusion-Generated Synthetic Images for Fair Image Classification

利用扩散生成的合成图像实现公平的图像分类

Abhipsa Basu, Aviral Gupta, Abhijnya Bhat, R. Venkatesh Babu

AI总结 本文提出利用扩散模型微调技术生成组平衡数据，以提升图像分类的公平性，实验表明其在去偏效果上优于现有方法。

Comments Accepted to AAAI AISI Track, 2026

Parameterized Approximation Algorithms for TSP on Non-Metric Graphs

参数化近似算法用于非度量图的旅行商问题

Jingyang Zhao, Zimo Sheng, Mingyu Xiao

AI总结 本文提出改进的参数化近似算法，针对一般图中违反三角不等式的顶点数p和使图变为度量图的最小顶点数q，分别获得1.5和3的近似比。

Comments A preliminary version of this article was presented at AAAI 2026

PSA-MF: Personality-Sentiment Aligned Multi-Level Fusion for Multimodal Sentiment Analysis

PSA-MF：基于人格-情感对齐的多级融合用于多模态情感分析

Heng Xie, Kang Zhu, Zhengqi Wen, Jianhua Tao, Xuefei Liu, Ruibo Fu, Changsheng Li

AI总结 PSA-MF通过引入人格-情感对齐和多级融合方法，提升多模态情感分析的识别性能。

Comments AAAI 2026 accepted

MDiff4STR: Mask Diffusion Model for Scene Text Recognition

MDiff4STR：用于场景文本识别的掩码扩散模型

Yongkun Du, Miaomiao Zhao, Songlin Fan, Zhineng Chen, Caiyan Jia, Yu-Gang Jiang

AI总结 MDiff4STR通过改进的掩码扩散模型在场景文本识别中实现更高的准确性和快速推断。

Comments Accepted by AAAI 2026 (Oral)

A Fast Heuristic Search Approach for Energy-Optimal Profile Routing for Electric Vehicles

一种用于电动车能耗最优路径规划的快速启发式搜索方法

Saman Ahmadi, Mahdi Jalili

AI总结 本文提出了一种基于多目标A*搜索的快速启发式搜索方法，用于电动车能耗最优路径规划，通过路径支配规则减少复杂路径处理，实现高效路径搜索。

Comments 13 pages, 5 Figures, 1 table, To appear as part of AAAI 2026 Proceedings

LLM-as-a-Judge for Scalable Test Coverage Evaluation: Accuracy, Operational Reliability, and Cost

基于LLM的判别器用于可扩展的测试覆盖率评估：准确性、操作可靠性与成本

Donghao Huang, Shila Chew, Anna Dutkiewicz, Zhaoxia Wang

机构 * School of Computing and Information Systems（计算与信息系统学院） ； Singapore Management University（新加坡管理大学） ； Research and Development（研发） 纠错

AI总结 LLM-as-a-Judge通过低成本高可靠性的测试覆盖率评估方法，显著降低评估成本并提升准确性。

Comments 7 pages, accepted by the AAAI 2026 Workshop on Next Gen Code Development with Collaborative AI Agents

DrawingBench: Evaluating Spatial Reasoning and UI Interaction Capabilities of Large Language Models through Mouse-Based Drawing Tasks

DrawingBench：通过基于鼠标绘制的任务评估大语言模型的空间推理和UI交互能力

Hyunjun Kim, Sooyoung Ryu

AI总结 DrawingBench通过基于鼠标绘制的任务评估大语言模型的空间推理和UI交互能力，揭示了模型在复杂场景中的表现及外部监督的重要性。

Comments AAAI 2026 TrustAgent Workshop

Adaptive Evidential Learning for Temporal-Semantic Robustness in Moment Retrieval

自适应证据学习用于时刻检索中的时序-语义鲁棒性

Haojian Huang, Kaijing Ma, Jin Chen, Haodong Chen, Zhou Wu, Xianghao Zang, Han Fang, Chao Ban, Hao Sun, Mulin Chen, Zhongjiang He

AI总结 本文提出DEMR框架，通过引入RFF模块和几何正则化器，提升时刻检索中对复杂视频场景的适应性与鲁棒性。

Comments Accepted by AAAI 2026, 10 pages, 9 figures, 5 tables

Hybrid-DMKG: A Hybrid Reasoning Framework over Dynamic Multimodal Knowledge Graphs for Multimodal Multihop QA with Knowledge Editing

Hybrid-DMKG: 一种基于动态多模态知识图谱的混合推理框架用于多模态多跳问答与知识编辑

Li Yuan, Qingfei Huang, Bingshan Zhu, Yi Cai, Qingbao Huang, Changmeng Zheng, Zikun Deng, Tao Wang

AI总结 Hybrid-DMKG通过动态多模态知识图谱实现多跳问答中的准确推理，提升对知识更新的鲁棒性。

Comments Accepted by AAAI 2026

PanFlow: Decoupled Motion Control for Panoramic Video Generation

PanFlow：全景视频生成的解耦运动控制

Cheng Zhang, Hanwen Liang, Donny Y. Chen, Qianyi Wu, Konstantinos N. Plataniotis, Camilo Cruz Gambardella, Jianfei Cai

AI总结 PanFlow通过解耦动态摄像机旋转与输入光学流，提升全景视频生成的运动控制精度和质量。

Comments Accepted by AAAI. Code: https://github.com/chengzhag/PanFlow

Augmented Runtime Collaboration for Self-Organizing Multi-Agent Systems: A Hybrid Bi-Criteria Routing Approach

增强型运行时协作用于自组织多智能体系统：一种混合双标准路由方法

Qingwen Yang, Feiyu Qu, Tiezheng Guo, Yanyi Liu, Yingyou Wen

AI总结 本文提出BiRouter，一种混合双标准路由方法，用于自组织多智能体系统，通过平衡重要性和连续性指标提升任务路由性能和鲁棒性。

Comments Accepted at The 40th Annual AAAI Conference on Artificial Intelligence

When Human Preferences Flip: An Instance-Dependent Robust Loss for RLHF

当人类偏好翻转时：面向RLHF的实例依赖性鲁棒损失

Yifan Xu, Xichen Ye, Yifan Chen, Qiaosheng Zhang

AI总结 本文提出了一种面向RLHF的实例依赖性鲁棒损失算法，通过建模偏好翻转机制和引入实例依赖的翻转概率，提升对齐算法的鲁棒性。

Comments Accepted by AAAI-26-AIA

AgentODRL: A Large Language Model-based Multi-agent System for ODRL Generation

AgentODRL: 基于大语言模型的多智能体系统用于ODRL生成

Wanle Zhong, Keman Huang, Xiaoyong Du

AI总结 AgentODRL通过多智能体系统和大语言模型技术，提升ODRL生成的准确性和效率。

Comments Accepted by AAAI 2026. 9 pages, 1 figure

SatireDecoder: Visual Cascaded Decoupling for Enhancing Satirical Image Comprehension

SatireDecoder: 视觉级联解耦以增强讽刺图像理解

Yue Jiang, Haiwei Xue, Minghao Han, Mingcheng Li, Xiaolu Hou, Dingkang Yang, Lihua Zhang, Xu Zheng

AI总结 SatireDecoder通过视觉级联解耦和不确定性分析策略，提升讽刺图像理解的准确性和语义层次。

Comments Accepted by AAAI 2026

Hyperbolic Continuous Structural Entropy for Hierarchical Clustering

双曲连续结构熵用于层次聚类

Guangjie Zeng, Hao Peng, Angsheng Li, Li Sun, Chunyang Liu, Shengze Li, Yicheng Pan, Philip S. Yu

AI总结 本文提出HypCSE模型，通过双曲空间映射和结构增强图优化，实现更高效的层次聚类。

Comments 14 pages, accepted by AAAI 2026

SCALE: Selective Resource Allocation for Overcoming Performance Bottlenecks in Mathematical Test-time Scaling

SCALE：面向数学测试时间扩展中性能瓶颈的有选择性资源分配

Yang Xiao, Chunpu Xu, Ruifeng Yuan, Jiashuo Wang, Wenjie Li, Pengfei Liu

AI总结 SCALE通过有选择性地分配计算资源，提升数学推理性能，显著优于均匀扩展方法。

Comments accepted by AAAI 2026

Whose Personae? Synthetic Persona Experiments in LLM Research and Pathways to Transparency

谁的人设？LLM研究中的人设实验及透明化路径

Jan Batzner, Volker Stocker, Bingjun Tang, Anusha Natarajan, Qinhao Chen, Stefan Schmid, Gjergji Kasneci

AI总结 本文探讨了LLM研究中合成人设实验的代表性与生态效度问题，提出透明化检查表以提升评估的严谨性和实证性。

Comments Published at AAAI/ACM AIES 2025. Presented at NeurIPS 2025 Workshop Evaluating the Evolving LLM Lifecycle: Benchmarks, Emergent Abilities, and Scaling

Journal ref Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 8(1), 2025, 343-354

PEOAT: Personalization-Guided Evolutionary Question Assembly for One-Shot Adaptive Testing

PEOAT：基于个性化引导的进化问题组装用于一次性自适应测试

Xiaoshan Yu, Ziwei Huang, Shangshang Yang, Ziwen Wang, Haiping Ma, Xingyi Zhang

AI总结 PEOAT通过个性化引导的进化算法，解决一次性自适应测试中高效选择最优题目的问题，提升测试效率和准确性。

Comments AAAI-2026, 9 pages

AerialMind: Towards Referring Multi-Object Tracking in UAV Scenarios

AerialMind：迈向无人机场景中的指称多目标跟踪

Chenglizhao Chen, Shaofeng Liang, Runwei Guan, Xiaolou Sun, Haocheng Zhao, Haiyun Jiang, Tao Huang, Henghui Ding, Qing-Long Han

AI总结 AerialMind是首个针对无人机场景的RMOT基准，通过COALA框架和HETrack方法提升无人机的自然语言交互能力与多目标跟踪性能。

Comments AAAI 2026

Probabilistic Hash Embeddings for Online Learning of Categorical Features

概率哈希嵌入用于在线学习分类特征

Aodong Li, Abishek Sankararaman, Balakrishnan Narayanaswamy

AI总结 本文提出概率哈希嵌入模型，用于在线学习中处理动态变化的分类特征，通过贝叶斯方法实现高效且内存高效的嵌入学习。

Comments AAAI 2026 Oral

CrossVid: A Comprehensive Benchmark for Evaluating Cross-Video Reasoning in Multimodal Large Language Models

CrossVid: 一个用于评估多模态大语言模型在跨视频推理中的综合基准

Jingyao Li, Jingyun Wang, Molin Tan, Haochen Wang, Cilin Yan, Likun Shi, Jiayin Cai, Xiaolong Jiang, Yao Hu

AI总结 CrossVid是首个用于评估多模态大语言模型跨视频推理能力的综合基准，通过多样化的任务和数据集验证了模型在复杂视频推理任务中的表现。

Comments Accepted to AAAI 2026 (main track). For code and data, see https://github.com/chuntianli666/CrossVid

Bridging Granularity Gaps: Hierarchical Semantic Learning for Cross-domain Few-shot Segmentation

弥合粒度差距：面向跨域少样本分割的层次语义学习

Sujun Sun, Haowen Gu, Cheng Xie, Yanxu Ren, Mingwu Ren, Haofeng Zhang

AI总结 本文提出层次语义学习框架，通过双风格随机化和层次语义挖掘模块，解决跨域少样本分割中粒度差距问题，提升语义区分能力。

Comments Accepted by AAAI 2026

CountSteer: Steering Attention for Object Counting in Diffusion Models

CountSteer：在扩散模型中通过引导注意力实现物体计数

Hyemin Boo, Hyoryung Kim, Myungjin Lee, Seunghyeon Lee, Jiyoung Lee, Jang-Hwan Choi, Hyunsoo Cho

AI总结 CountSteer通过引导扩散模型的注意力机制，提升指定物体数量的生成精度，实现更可控的图像生成。

Comments Accepted to AAAI 2026 Workshop on Shaping Responsible Synthetic Data in the Era of Foundation Models (RSD)

"As Eastern Powers, I will veto." : An Investigation of Nation-level Bias of Large Language Models in International Relations

作为东方国家，我将 veto。：对大型语言模型在国际关系领域国家层面偏见的调查

Jonghyeon Choi, Yeonjun Choi, Hyun-chul Kim, Beakcheol Jang

AI总结 本文研究了大型语言模型在国际关系领域中的国家层面偏见，提出了一种去偏框架以减少偏见并提升性能。

Comments 21 pages, 4 figures. This is the extended version of the paper accepted at AAAI 2026, which includes all technical appendices and additional experimental details

Journal ref AAAI 2026

LiNeXt: Revisiting LiDAR Completion with Efficient Non-Diffusion Architectures

LiNeXt：重新审视基于高效非扩散架构的LiDAR补全

Wenzhe He, Xiaojun Chen, Ruiqi Wang, Ruihui Li, Huilong Pi, Jiapeng Zhang, Zhuo Tang, Kenli Li

AI总结 LiNeXt通过高效非扩散架构实现快速准确的LiDAR点云补全，提升实时性能并减少计算开销。

Comments 18 pages, 13 figures, Accepted to AAAI 2026

Multivariate Gaussian Representation Learning for Medical Action Evaluation

多变量高斯表示学习用于医学动作评估

Luming Yang, Haoxian Liu, Siqing Li, Alper Yilmaz

AI总结 本文提出GaussMedAct框架，通过多变量高斯表示学习提升医学动作分析的准确率和鲁棒性。

Comments Accepted to AAAI 2026