AAAI

TOSC: Task-Oriented Shape Completion for Open-World Dexterous Grasp Generation from Partial Point Clouds

TOSC：面向任务的形状补全用于从部分点云生成开放世界灵巧抓取

Weishang Wu, Yifei Shi, Zhiping Cai

AI总结 本文提出面向任务的形状补全方法，通过生成任务导向的接触区域完成方案，提升开放世界中灵巧抓取的性能，改进抓取位移和 Chamfer 距离。

Comments Accepted to AAAI 2026

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 40(13), 10781-10789 (2026)

Incremental Data-Driven Policy Synthesis via Game Abstractions

通过游戏抽象实现增量数据驱动的策略合成

Irmak Sağlam, Mahdi Nazeri, Alessandro Abate, Sadegh Soudjani, Anne-Kathrin Schmuck

AI总结 本文提出一种基于数据驱动和抽象的控制框架，通过增量游戏求解方法，在系统动态数据积累时逐步构建抽象游戏图、获胜区域和控制策略，实现对未知离散时间随机动态系统的控制策略合成。

Comments Presented at the 40th Annual AAAI Conference on Artificial Intelligence AAAI'26 (Oral)

ExPath: Targeted Pathway Inference for Biological Knowledge Bases via Graph Learning and Explanation

ExPath：通过图学习和解释进行生物知识库的靶向通路推断

Rikuto Kotoge, Ziwei Yang, Zheng Chen, Yushun Dong, Yasuko Matsubara, Jimeng Sun, Yasushi Sakurai

AI总结 本文提出ExPath框架，通过图学习和解释技术，整合实验数据推断生物网络中的靶向通路，实验表明其在Fidelity+和Fidelity-指标上优于基线方法。

Comments Accepted at AAAI 2026 (Main Technical Track)

From Decision Trees to Boolean Logic: A Fast and Unified SHAP Algorithm

从决策树到布尔逻辑：一种快速且统一的SHAP算法

Alexander Nadel, Ron Wettenstein

AI总结 本文提出WOODELF算法，结合决策树、博弈论和布尔逻辑，实现快速统一的SHAP计算，支持CPU和GPU高效运行，显著提升大规模数据处理速度。

Comments Published at AAAI 2026

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 40, No. 29, 2026

Do LLMs Feel? Teaching Emotion Recognition with Prompts, Retrieval, and Curriculum Learning

LLMs能否感知情绪？通过提示、检索和课程学习进行情绪识别

Xinran Li, Yu Liu, Jiaqi Qiao, Xiujuan Xu

AI总结 本文提出PRC-Emo框架，结合提示工程、示范检索和课程学习，探索LLM在对话中感知情绪的能力，实验表明在IEMOCAP和MELD数据集上达到新的SOTA性能。

Comments Accepted at AAAI 2026

Journal ref Proc. AAAI Conf. on Artificial Intelligence, Vol. 40, No. 38, pp. 31778-31786 (2026)

How Wide and How Deep? Mitigating Over-Squashing of GNNs via Channel Capacity Constrained Estimation

多宽多深？通过通道容量约束估计缓解图神经网络的过压缩

Zinuo You, Jin Zheng, John Cartlidge

AI总结 本文提出C3E框架，通过信息论将隐藏维度和深度的选择建模为非线性优化问题，缓解图神经网络的过压缩问题，提升表示学习性能。

Comments 29 pages, 11 figures. Author manuscript accepted for the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26), January 2026

Journal ref AAAI 2026, Proceedings of the AAAI Conference on Artificial Intelligence, 40(33), 27890-27898

Poisoning with A Pill: Circumventing Detection in Federated Learning

用一粒药丸进行毒害：在联邦学习中规避检测

Hanxi Guo, Hao Wang, Tao Song, Tianhang Zheng, Yang Hua, Haibing Guan, Xiangyu Zhang

机构 * Purdue University（普渡大学） 纠错

AI总结 本文提出一种通用的攻击无关增强方法，通过在联邦学习训练中构造、生成并注入毒害（由现有攻击生成）到一个药丸（一种新型子网络结构）中，以提升现有毒害攻击的隐蔽性和有效性，揭示现有防御的不足。

Comments Accepted by AAAI 2026

LLMs for Game Theory: Entropy-Guided In-Context Learning and Adaptive CoT Reasoning

大型语言模型用于博弈论：基于熵的上下文学习与自适应推理

Tommaso Felice Banfi, Sashenka Gamage

AI总结 本文提出一种基于LLM的框架，通过熵引导的链式推理和自适应上下文检索，提升离散博弈任务的推理能力，实验显示其显著提高决策质量。

Comments Published at the AAAI 2026 Bridge: Logical and Symbolic Reasoning in Language Models (OpenReview)

End-to-end Contrastive Language-Speech Pretraining Model For Long-form Spoken Question Answering

端到端对比语言-语音预训练模型用于长形式语音问答

Jiliang Hu, Zuchao Li, Baoyuan Qi, Liu Guoming, Ping Wang

AI总结 本文提出CLSR模型，通过将音频特征转换为文本表示，提升长音频问答任务的性能，实验表明其优于现有方法。

Comments 12 pages, 7 figures, accepted by AAAI 2026

ChatCLIDS: Simulating Persuasive AI Dialogues to Promote Closed-Loop Insulin Adoption in Type 1 Diabetes Care

ChatCLIDS: 模拟说服性人工智能对话以促进1型糖尿病护理中的闭环胰岛素采用

Zonghai Yao, Talha Chafekar, Junda Wang, Shuo Han, Feiyun Ouyang, Junhui Qian, Lingxi Li, Hong Yu

AI总结 本文提出ChatCLIDS基准，用于评估基于LLM的说服对话在健康行为改变中的效果，通过模拟多轮交互和长期咨询场景，揭示当前LLM在行为改变中的局限性。

Comments Equal contribution for the first two authors. To appear in AAAI 2026 Special Track on AI for Social Impact

Journal ref AAAI 2026