AAAI

LiteGE: Lightweight Geodesic Embedding for Efficient Geodesics Computation and Non-Isometric Shape Correspondence

LiteGE：轻量级测地嵌入用于高效测地线计算和非等距形状对应

Yohanes Yudhi Adikusuma, Qixing Huang, Ying He

AI总结 LiteGE通过PCA处理信息体素的无符号距离场样本，构建紧凑且类别感知的形状描述符，实现高效测地线计算和非等距形状对应，显著降低内存使用和推理时间。

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

iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification

iSeal：用于可靠LLM所有权验证的加密指纹法

Zixun Xiong, Gaoyi Wu, Qingyang Yu, Mingyu Derek Ma, Lingfeng Yao, Miao Pan, Xiaojiang Du, Hao Wang

AI总结 iSeal通过在模型和外部模块中注入独特特征，结合纠错机制和相似性验证策略，实现对模型窃贼全程控制下的可靠LLM所有权验证，有效抵御指纹反学习和响应操控等攻击。

Comments Accepted by AAAI 2026

Journal ref Proc. AAAI Conf. Artif. Intell. 40(42): 23984-23992, 2026

Otter: Mitigating Background Distractions of Wide-Angle Few-Shot Action Recognition with Enhanced RWKV

Otter: 通过增强的RWKV缓解宽角少样本动作识别中的背景干扰

Wenbo Huang, Jinghui Zhang, Zhenghao Chen, Guang Li, Lei Zhang, Yang Cao, Fang Dong, Takahiro Ogawa, Miki Haseyama

AI总结 本文提出Otter，通过Compound Segmentation和Temporal Reconstruction模块提升宽角少样本动作识别性能，有效突出主体并重建时间关系，实验表明其在多个数据集上达到最优效果。

Comments Accepted by AAAI 2026 Oral

The Geometry of Dialogue: Graphing Language Models to Reveal Synergistic Teams for Multi-Agent Collaboration

对话的几何学：基于图谱的语言模型揭示多智能体协作的协同团队

Kotaro Furuya, Yuichi Kitagawa

AI总结 本文提出一种基于交互的自动团队组建框架，通过构建语言模型图谱揭示多智能体协作的协同团队，实验表明其能发现功能一致的团队并优于随机基线。

Comments Accepted at the AAAI-26 Workshop on LLM-based Multi-Agent Systems: Towards Responsible, Reliable, and Scalable Agentic Systems (LaMAS 2026) as an oral presentation

From Detection to Diagnosis: Advancing Hallucination Analysis with Automated Data Synthesis

从检测到诊断：通过自动化数据合成推进幻觉分析

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

AI总结 本文提出从检测到诊断的新范式，通过自动化数据合成生成高质量训练样本，训练出HDM-4B-RL模型，在幻觉诊断任务中超越现有检测模型，实现更可靠的生成式AI系统。

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

Do Language Models Associate Sound with Meaning? A Multimodal Study of Sound Symbolism

语言模型是否将声音与意义关联？一种多模态研究声音象征主义

Jinhong Jeong, Sunghyun Lee, Jaeyoung Lee, Seonah Han, Youngjae Yu

机构 * Yonsei University（延世大学） ； Seoul National University（首尔国立大学） ； Korea University（韩国大学） 纠错

AI总结 研究语言模型在多模态输入中对声音象征性的处理，通过分析不同语义维度下的音节注意力分数，揭示模型对声音与意义关联的理解机制。

Comments 33 pages, 27 tables, 10 figures, accepted to AAAI 2026 (Oral)

Flexible Concept Bottleneck Model

灵活的概念瓶颈模型

Xingbo Du, Qiantong Dou, Lei Fan, Rui Zhang

AI总结 本文提出灵活的概念瓶颈模型（FCBM），通过动态概念适应和改进的稀疏max模块，提升模型在新概念下的适应性与灵活性，实验表明其在多个基准上的表现优异。

Comments To appear in AAAI 2026

StoryBox: Collaborative Multi-Agent Simulation for Hybrid Bottom-Up Long-Form Story Generation Using Large Language Models

StoryBox：基于大型语言模型的混合自底向上长篇故事生成的协作多智能体模拟

Zehao Chen, Rong Pan, Haoran Li

AI总结 本文提出混合自底向上长篇故事生成方法，利用多智能体模拟生成超过10000字的连贯故事，通过智能体交互产生有机情节发展，优于传统自顶向下的结构化方法。

Comments Accepted by AAAI 2026. Project: https://storyboxproject.github.io

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 2026, 40(36), 30359-30367

SVGBuilder: Component-Based Colored SVG Generation with Text-Guided Autoregressive Transformers

SVGBuilder：基于组件的带颜色SVG生成与文本引导自回归变换

Zehao Chen, Rong Pan

AI总结 SVGBuilder基于组件的文本引导自回归变换生成高质量彩色SVG，显著降低计算开销，提升效率，引入ColorSVG-100K数据集提升模型多样性与颜色信息。

Comments Accepted by AAAI 2025. Project: https://svgbuilder.github.io

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 2025, 39(3), 2358-2366