AAAI

T-LoRA: Single Image Diffusion Model Customization Without Overfitting

T-LoRA: 单张图像扩散模型定制化无需过拟合

Vera Soboleva, Aibek Alanov, Andrey Kuznetsov, Konstantin Sobolev

AI总结 T-LoRA通过时间步依赖的低秩适应框架实现单张图像扩散模型定制化，有效避免过拟合，提升概念保真度与文本对齐的平衡。

Comments AAAI 2026

Benchmarking LLMs for Political Science: A United Nations Perspective

对政治科学的LLM进行基准测试：一个联合国视角

Yueqing Liang, Liangwei Yang, Chen Wang, Congying Xia, Rui Meng, Xiongxiao Xu, Haoran Wang, Ali Payani, Kai Shu

机构 * Google Cloud AI Research（谷歌云人工智能研究） 纠错

AI总结 本文提出联合国基准（UNBench），用于评估LLMs在政治科学中的能力，涵盖联合国决策过程的三个阶段。

Comments This paper has been accepted at AAAI 2026 as an oral paper

VitalDiagnosis: AI-Driven Ecosystem for 24/7 Vital Monitoring and Chronic Disease Management

VitalDiagnosis：基于AI的生态系统，用于24/7生命体征监测和慢性病管理

Zhikai Xue, Tianqianjin Lin, Pengwei Yan, Ruichun Wang, Yuxin Liu, Zhuoren Jiang, Xiaozhong Liu

AI总结 VitalDiagnosis利用AI技术，通过整合可穿戴设备数据与大语言模型，实现24/7生命体征监测和慢性病管理的主动参与，提升患者自我管理能力并减少临床工作量。

Comments Accepted by AAAI 2026 Demo

Language Models Entangle Language and Culture

语言模型融合语言与文化

Shourya Jain, Paras Chopra

机构 * Lossfunk 纠错

AI总结 研究发现语言选择影响LLM生成答案的文化上下文，导致低资源语言回答质量较低。

Comments Accepted at LM4UC Workshop at AAAI'26, Submitted to ACL 2026. 17 pages, 7 figures

Heterogeneous Uncertainty-Guided Composed Image Retrieval with Fine-Grained Probabilistic Learning

异质不确定性引导的复合图像检索与细粒度概率学习

Haomiao Tang, Jinpeng Wang, Minyi Zhao, Guanghao Meng, Ruisheng Luo, Long Chen, Shu-Tao Xia

AI总结 本文提出异质不确定性引导范式，通过细粒度概率学习提升复合图像检索的鲁棒性和准确性。

Comments Accepted for publication and oral presentation at AAAI 2026

Radiation-Preserving Selective Imaging for Pediatric Hip Dysplasia: A Cross-Modal Ultrasound-Xray Policy with Limited Labels

辐射保护的儿童髋关节发育不良选择性成像：一种跨模态超声-X射线策略（有限标注）

Duncan Stothers, Ben Stothers, Emily Schaeffer, Kishore Mulpuri

AI总结 本文提出了一种跨模态超声-X光策略，通过有限标注实现儿童髋关节发育不良的可解释测量和选择性成像。

Comments Accepted (with oral presentation) to the AAAI 2026 AI for Medicine and Healthcare Bridge Program Awarded Best Paper Runner-Up at the AAAI 2026 AI for Medicine and Healthcare Bridge Program

Divide, Conquer and Unite: Hierarchical Style-Recalibrated Prototype Alignment for Federated Medical Segmentation

分割、征服与联合：基于层次化风格校正原型对齐的联邦医学分割

Xingyue Zhao, Wenke Huang, Xingguang Wang, Haoyu Zhao, Linghao Zhuang, Anwen Jiang, Guancheng Wan, Mang Ye

AI总结 FedBCS通过层次化风格校正原型对齐方法，解决联邦学习中医学分割的特征异质性问题，提升模型鲁棒性与分割精度。

Comments Accepted at AAAI-26

DECOR: Deep Embedding Clustering with Orientation Robustness

DECOR：具有方向鲁棒性的深度嵌入聚类

Fiona Victoria Stanley Jothiraj, Arunaggiri Pandian Karunanidhi, Seth A. Eichmeyer

机构 * Micron Technology（micron技术公司） 纠错

AI总结 DECOR通过引入方向鲁棒性，实现了在复杂、不平衡的晶圆数据中可靠且高效的聚类，提升自动化视觉检测系统的性能。

Comments Accepted to the KGML Bridge at AAAI 2026 (non-archival)

Modelling the Effects of Hearing Loss on Neural Coding in the Auditory Midbrain with Variational Conditioning

利用变分条件建模听觉脑干中听觉损失对神经编码的影响

Lloyd Pellatt, Fotios Drakopoulos, Shievanie Sabesan, Nicholas A. Lesica

AI总结 本文提出变分条件模型，直接从听觉中脑神经活动记录中学习听觉损失的空间编码，准确预测神经响应方差并模拟真实活动，为未来参数化听觉损失补偿模型提供基础。

Comments 12 pages, 3 figures, presented at AAAI 2026

Embracing Ambiguity: Bayesian Nonparametrics and Stakeholder Participation for Ambiguity-Aware Safety Evaluation

拥抱模糊性：基于贝叶斯非参数方法和利益相关者参与的模糊性感知安全评估

Yanan Long

AI总结 本文提出了一种基于贝叶斯非参数方法和利益相关者参与的框架，用于模糊性感知的安全评估，通过量化尾部风险和整合利益相关者偏好提升生成模型的可信度。

Comments AAAI 2026 workshop MURE

On the Exponential Convergence for Offline RLHF with Pairwise Comparisons

关于通过成对比较进行离线RLHF的指数收敛性

Zhirui Chen, Vincent Y. F. Tan

AI总结 本文提出了一种在离线RLHF中通过成对比较实现指数收敛的算法RL-LOW，并推导了实例依赖的下界。

Comments Accepted as an oral presentation at AAAI 2026 (AI Alignment Track)

Vehicle Routing with Finite Time Horizon using Deep Reinforcement Learning with Improved Network Embedding

具有有限时间跨度的车辆路径问题使用改进的网络嵌入深度强化学习

Ayan Maity, Sudeshna Sarkar

AI总结 本文提出了一种改进的网络嵌入深度强化学习方法，用于解决具有有限时间跨度的车辆路径问题，通过结合局部节点嵌入和全局图表示，提升了客户服务水平和求解效率。

Comments Accepted at AAAI-26 Workshop on AI for Urban Planning

Pb4U-GNet: Resolution-Adaptive Garment Simulation via Propagation-before-Update Graph Network

Pb4U-GNet：通过传播前更新图网络实现分辨率自适应的服装模拟

Aoran Liu, Kun Hu, Clinton Ansun Mo, Qiuxia Wu, Wenxiong Kang, Zhiyong Wang

AI总结 Pb4U-GNet通过动态传播深度控制和几何感知更新缩放，实现了在不同分辨率网格上的自适应服装模拟。

Comments Camera-ready version accepted at AAAI 2026

LiViBench: An Omnimodal Benchmark for Interactive Livestream Video Understanding

LiViBench：面向交互式直播视频理解的多模态基准测试

Xiaodong Wang, Langling Huang, Zhirong Wu, Xu Zhao, Teng Xu, Xuhong Xia, Peixi Peng

AI总结 LiViBench是首个面向交互式直播视频的多模态基准测试，通过定制化两阶段指令微调和视频到评论检索模块，提升模型对直播视频的理解能力，并在多个基准测试中取得优异成绩。

Comments AAAI 2026 Main Track

Adaptive Fidelity Estimation for Quantum Programs with Graph-Guided Noise Awareness

具有图引导噪声意识的量子程序适应保真度估计

Tingting Li, Ziming Zhao, Jianwei Yin

AI总结 QuFid通过图引导噪声意识框架，实现量子程序的自适应保真度估计，有效降低测量成本并保持保真度精度。

Comments Published in AAAI 2026;

SOSControl: Enhancing Human Motion Generation through Saliency-Aware Symbolic Orientation and Timing Control

SOSControl: 通过具有显著性意识的符号性方向和时间控制增强人体运动生成

Ho Yin Au, Junkun Jiang, Jie Chen

AI总结 SOSControl通过显著性意识的符号性方向和时间控制提升人体运动生成的质量与可控性。

Comments Accepted by AAAI 2026

Does Less Hallucination Mean Less Creativity? An Empirical Investigation in LLMs

减少幻觉是否意味着减少创造力？在大语言模型中的实证研究

Mohor Banerjee, Nadya Yuki Wangsajaya, Syed Ali Redha Alsagoff, Min Sen Tan, Zachary Choy Kit Chun, Alvin Chan Guo Wei

AI总结 本研究探讨了三种减少幻觉的方法对大语言模型创造力的影响，发现CoVe促进发散思维，DoLa抑制创造力，RAG影响较小，为科学应用中的准确性与创造性平衡提供指导。

Comments Accepted at the AAAI 2026 Workshop on AI for Scientific Research (AI4Research)

Beyond Boundaries: Leveraging Vision Foundation Models for Source-Free Object Detection

超越边界：利用视觉基础模型进行无源目标检测

Huizai Yao, Sicheng Zhao, Pengteng Li, Yi Cui, Shuo Lu, Weiyu Guo, Yunfan Lu, Yijie Xu, Hui Xiong

AI总结 本文提出一种利用视觉基础模型提升无源目标检测性能的新框架，通过增强特征对齐和标签质量，实现跨领域迁移和辨别性提升。

Comments Accepted to AAAI 2026. Extended version with full Appendix

Explaining Tournament Solutions with Minimal Supports

用最小支持解释竞赛解

Clément Contet, Umberto Grandi, Jérôme Mengin

AI总结 本文研究了如何通过最小支持来解释竞赛解，提出了多项式时间算法计算各竞赛规则下的最小支持，并展示了其在生成认证解释中的应用。

Comments This paper is the extended version of Contet, Grandi, and Mengin. 2026. Explaining Tournament Solutions with Minimal Supports. In Proceedings of the 40th AAAI Conference on Artificial Intelligence

BayesAgent: Bayesian Agentic Reasoning Under Uncertainty via Verbalized Probabilistic Graphical Modeling

BayesAgent: 通过 verbalized 概率图模型实现不确定性下的贝叶斯代理推理

Hengguan Huang, Xing Shen, Songtao Wang, Lingfa Meng, Dianbo Liu, David Alejandro Duchene, Hao Wang, Samir Bhatt

AI总结 BayesAgent 通过 verbalized 概率图模型实现不确定性下的贝叶斯代理推理，提升置信度校准和文本生成质量。

Comments Accepted to AAAI 2026

GOMPSNR: Reflourish the Signal-to-Noise Ratio Metric for Audio Generation Tasks

GOMPSNR：为音频生成任务复兴信号噪声比度量

Lingling Dai, Andong Li, Cheng Chi, Yifan Liang, Xiaodong Li, Chengshi Zheng

AI总结 本文提出GOMPSNR，通过引入相位距离项改进SNR，提升音频生成任务中客观度量的可靠性，并通过两种新型损失函数优化模型性能。

Comments Accepted by AAAI 2026

GeoSteer: Faithful Chain-of-Thought Steering via Latent Manifold Gradients

GeoSteer: 通过潜在流形梯度实现忠实的思维链引导

Kentaro Kazama, Daiki Shirafuji, Tatsuhiko Saito

AI总结 GeoSteer通过学习高质量CoT轨迹的低维流形，利用梯度引导提升LLM中间推理质量，实验显示在GSM8k数据集上准确率和推理质量均显著提升。

Comments The Third workshop of NeusymBridge @AAAI 2026 (Bridging Neurons and Symbols for NLP and Knowledge Graph Reasoning)

OceanSplat: Object-aware Gaussian Splatting with Trinocular View Consistency for Underwater Scene Reconstruction

OceanSplat: 基于三目视图一致性的物体感知高斯点云法用于水下场景重建

Minseong Kweon, Jinsun Park

AI总结 OceanSplat通过三目视图一致性与深度感知调整，实现高保真水下场景重建与恢复。

Comments Accepted to AAAI 2026. Project page: https://oceansplat.github.io

ConSurv: Multimodal Continual Learning for Survival Analysis

ConSurv：面向生存分析的多模态持续学习

Dianzhi Yu, Conghao Xiong, Yankai Chen, Wenqian Cui, Xinni Zhang, Yifei Zhang, Hao Chen, Joseph J. Y. Sung, Irwin King

AI总结 ConSurv是首个用于生存分析的多模态持续学习方法，通过多阶段混合专家和特征受限重放技术，有效解决灾难性遗忘和复杂模态交互问题。

Comments 14 pages, 4 figures. This is the extended version of the paper accepted at AAAI 2026, which includes all technical appendices and additional experimental details

SurgPub-Video: A Comprehensive Surgical Video Dataset for Enhanced Surgical Intelligence in Vision-Language Model

SurgPub-Video: 一个全面的外科视频数据集，用于增强视觉-语言模型中的外科智能

Yaoqian Li, Xikai Yang, Dunyuan Xu, Yang Yu, Litao Zhao, Xiaowei Hu, Jinpeng Li, Pheng-Ann Heng

AI总结 SurgPub-Video数据集和SurgLLaVA-Video模型通过提供高质量外科视频数据和专门的视觉-语言模型，提升了手术场景分析的智能水平。

Journal ref AAAI-2026

Adaptive Riemannian Graph Neural Networks

自适应黎曼图神经网络

Xudong Wang, Chris Ding, Tongxin Li, Jicong Fan

机构 * AAAI-2026 纠错

AI总结 自适应黎曼图神经网络通过学习连续各向异性度量张量场，有效捕捉图数据的复杂几何异质性，提升在同质和异质数据集上的性能。

Comments Accepted in The Fortieth AAAI Conference on Artificial Intelligence (AAAI-26), Main Technical Track

HCF: Hierarchical Cascade Framework for Distributed Multi-Stage Image Compression

HCF：分布式多阶段图像压缩的分层级联框架

Junhao Cai, Taegun An, Chengjun Jin, Sung Il Choi, Juhyun Park, Changhee Joo

AI总结 HCF通过分层级联框架实现分布式多阶段图像压缩的高码率失真性能和高效计算，优于现有方法在PSNR和BD-Rate上的表现。

Comments Accepted at AAAI 2026 as a Conference Paper (Oral Presentation)

The Case for "Thick Evaluations" of Cultural Representation in AI

文化表征在AI中的‘厚评价’案例

Rida Qadri, Mark Diaz, Ding Wang, Michael Madaio

AI总结 本文提出‘厚评估’框架，通过南亚研讨会研究社区对AI生成文化图像的解读方式，以更细致的测量方法评估AI输出中的文化表征。

Comments 10 pages

Journal ref Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 8(3), 2067-2080 (2025)

Object-Centric Latent Action Learning

基于对象的潜在动作学习

Albina Klepach, Alexander Nikulin, Ilya Zisman, Denis Tarasov, Alexander Derevyagin, Andrei Polubarov, Nikita Lyubaykin, Igor Kiselev, Vladislav Kurenkov

AI总结 本文提出了一种基于对象的潜在动作学习框架，通过自监督对象中心预训练解耦代理与干扰背景动态，提升动作标签的鲁棒性，从而改善模仿学习和代理适应效率。

Comments Accepted by AAAI 2026 (Oral). Source code: https://github.com/dunnolab/object-centric-lapo

Membership Inference Test: Auditing Training Data in Object Classification Models

成员推断测试：在目标分类模型中审计训练数据

Gonzalo Mancera, Daniel DeAlcala, Aythami Morales, Ruben Tolosana, Julian Fierrez

AI总结 本研究提出适用于目标分类模型的MINT架构，通过分析训练数据使用情况，实现70%-80%的识别精度，提升数据透明度。

Comments Deployable AI (DAI 2025) workshop co-located with AAAI-25