AAAI

Aggregating Diverse Cue Experts for AI-Generated Image Detection

聚合多样化线索专家用于AI生成图像检测

Lei Tan, Shuwei Li, Mohan Kankanhalli, Robby T. Tan

AI总结 本文提出MCAN框架，通过整合多种互补线索提升AI生成图像检测的泛化能力，实验显示在GenImage数据集上性能优于现有方法。

Comments Accepted by AAAI 2026

Evaluating the Ability of Explanations to Disambiguate Models in a Rashomon Set

评估解释在拉索莫集合中区分模型的能力

Kaivalya Rawal, Eoin Delaney, Zihao Fu, Sandra Wachter, Chris Russell

AI总结 本文提出AXE方法，用于评估模型解释的质量，以区分Rashomon集合中模型的行为差异，防止虚假解释误导评估。

Comments This is a preprint of the paper published at the MURE workshop, AAAI 2026, which builds on a preprint of separate work published at FAccT 2025 (arXiv:2505.10399)

All Required, In Order: Phase-Level Evaluation for AI-Human Dialogue in Healthcare and Beyond

全部必要，按顺序：面向AI-人类对话在医疗保健及其他领域的相级评估

Shubham Kulkarni, Alexander Lyzhov, Shiva Chaitanya, Preetam Joshi

AI总结 OIP-SCE是一种用于评估AI-人类对话在医疗及其他领域中合规性的方法，通过相级证据确保临床义务按顺序满足，提升AI与临床流程的对齐度。

Comments Accepted at the AI for Medicine and Healthcare (AIMedHealth) Bridge Program, AAAI-26, Singapore. Full-length paper; to appear in Proceedings of Machine Learning Research (PMLR)

M$^2$FMoE: Multi-Resolution Multi-View Frequency Mixture-of-Experts for Extreme-Adaptive Time Series Forecasting

M$^2$FMoE：多分辨率多视角频率混合专家用于极端适应时间序列预测

Yaohui Huang, Runmin Zou, Yun Wang, Laeeq Aslam, Ruipeng Dong

AI总结 M$^2$FMoE通过多分辨率多视角频率混合专家模型，有效捕捉时间序列中的常规和极端模式，提升极端事件适应性预测性能。

Comments Accepted by AAAI 2026

CtrlFuse: Mask-Prompt Guided Controllable Infrared and Visible Image Fusion

CtrlFuse: 基于掩码提示的可控红外与可见图像融合

Yiming Sun, Yuan Ruan, Qinghua Hu, Pengfei Zhu

AI总结 CtrlFuse通过基于掩码提示的可控融合框架，实现红外与可见图像的交互式动态融合，提升任务性能与融合质量。

Comments 18 pages,22 figures,published to AAAI 2026

How Hard Is It to Rig a Tournament When Few Players Can Beat or Be Beaten by the Favorite?

当少数玩家能击败或被 favorites 击败时，如何操纵比赛？

Zhonghao Wang, Junqiang Peng, Yuxi Liu, Mingyu Xiao

AI总结 本文研究了如何在少数玩家能击败或被 favorites 击败时操纵比赛，证明了当参数化为 v* 的入度或出度时，Tournament Fixing 问题可高效解决。

Comments Accepted by AAAI 2026

DiffMM: Efficient Method for Accurate Noisy and Sparse Trajectory Map Matching via One Step Diffusion

DiffMM: 一种用于准确匹配噪声和稀疏轨迹图的高效方法通过一步扩散

Chenxu Han, Sean Bin Yang, Jilin Hu

AI总结 DiffMM通过一步扩散方法，高效准确地匹配噪声和稀疏轨迹，提升交通调度和流分析的应用性能。

Comments AAAI-26

SUMMPILOT: Bridging Efficiency and Customization for Interactive Summarization System

SUMMPILOT：在交互式摘要系统中平衡效率与定制化

JungMin Yun, Juhwan Choi, Kyohoon Jin, Soojin Jang, Jinhee Jang, YoungBin Kim

AI总结 SUMMPILOT是一种通过交互式组件实现个性化摘要的系统，结合大型语言模型提升效率与定制化能力。

Comments Accepted to AAAI 2025 Demonstration Track

M3SR: Multi-Scale Multi-Perceptual Mamba for Efficient Spectral Reconstruction

M3SR：多尺度多感知Mamba用于高效光谱重建

Yuze Zhang, Lingjie Li, Qiuzhen Lin, Zhong Ming, Fei Yu, Victor C. M. Leung

AI总结 M3SR通过多尺度多感知架构提升超光谱图像重建的准确性和效率，同时降低计算成本。

Comments Accepted by AAAI 2026

A New Strategy for Verifying Reach-Avoid Specifications in Neural Feedback Systems

神经反馈系统中验证可达-回避规范的新策略

Samuel I. Akinwande, Sydney M. Katz, Mykel J. Kochenderfer, Clark Barrett

机构 * Department of Aeronautics and Astronautics, Stanford University, Stanford, USA（航空与航天系，斯坦福大学） ； Department of Computer Science, Stanford University, Stanford, USA（计算机科学系，斯坦福大学） 纠错

AI总结 本文提出了一种新的算法，用于计算神经反馈系统中可达-回避规范的上界和下界近似值，并结合前向分析技术，构建统一的验证框架。

Comments Accepted to AAAI-2026 Bridge Program B10: Making Embodied AI Reliable with Testing and Formal Verification

Transformer-Based Approach for Automated Functional Group Replacement in Chemical Compounds

基于变压器的方法用于化学化合物中的功能基团替换自动化

Bo Pan, Zhiping Zhang, Kevin Spiekermann, Tianchi Chen, Xiang Yu, Liying Zhang, Liang Zhao

AI总结 本文提出了一种基于变压器的两阶段模型，用于自动化化学化合物中功能基团的替换，通过序列生成确保子结构修改，有效捕捉转换规则并提升生成化学结构的多样性与有效性。

Comments The 2nd AAAI Workshop on Foundation Models for Biological Discoveries at AAAI 2025

Multiplicative Orthogonal Sequential Editing for Language Models

乘法正交序列编辑用于语言模型

Hao-Xiang Xu, Jun-Yu Ma, Ziqi Peng, Yuhao Sun, Zhen-Hua Ling, Jia-Chen Gu

AI总结 MOSE通过乘法正交矩阵编辑方法提升语言模型的序列编辑性能，同时保持一般能力。

Comments Accepted by AAAI 2026

Ground What You See: Hallucination-Resistant MLLMs via Caption Feedback, Diversity-Aware Sampling, and Conflict Regularization

在所见之地接地：通过标题反馈、多样性感知采样和冲突正则化实现抗幻觉的MLLMs

Miao Pan, Wangjie Gan, Jintao Chen, Wenqi Zhang, Bing Sun, Jianwei Yin, Xuhong Zhang

AI总结 本文提出抗幻觉的MLLMs方法，通过标题反馈、多样性感知采样和冲突正则化减少幻觉，提升推理准确性。

Comments AAAI-2026 Poster

Backdoor Attacks on Open Vocabulary Object Detectors via Multi-Modal Prompt Tuning

通过多模态提示调优对开放词汇目标检测器进行后门攻击

Ankita Raj, Chetan Arora

AI总结 TrAP通过多模态提示调优对开放词汇目标检测器实施后门攻击，利用轻量级提示标记植入恶意行为，提升攻击成功率并改进下游任务性能。

Comments Accepted to AAAI 2026

PINGS-X: Physics-Informed Normalized Gaussian Splatting with Axes Alignment for Efficient Super-Resolution of 4D Flow MRI

PINGS-X: 基于轴对齐的物理信息归一化高斯点云的4D流体MRI超分辨率方法

Sun Jo, Seok Young Hong, JinHyun Kim, Seungmin Kang, Ahjin Choi, Don-Gwan An, Simon Song, Je Hyeong Hong

AI总结 PINGS-X通过轴对齐高斯点云和归一化方法提升4D流体MRI超分辨率效率与精度。

Comments Accepted at AAAI 2026. Supplementary material included after references. 27 pages, 21 figures, 11 tables

SpatialActor: Exploring Disentangled Spatial Representations for Robust Robotic Manipulation

SpatialActor: 探索解耦的空间表示以实现鲁棒的机器人操作

Hao Shi, Bin Xie, Yingfei Liu, Yang Yue, Tiancai Wang, Haoqiang Fan, Xiangyu Zhang, Gao Huang

机构 * Dexmal 纠错

AI总结 SpatialActor通过解耦语义和几何，提升机器人操作的鲁棒性和精确性，实现高准确率和强泛化能力。

Comments AAAI 2026 Oral | Project Page: https://shihao1895.github.io/SpatialActor

Format as a Prior: Quantifying and Analyzing Bias in LLMs for Heterogeneous Data

以格式为先：量化和分析LLM在异构数据中的偏见

Jiacheng Liu, Mayi Xu, Qiankun Pi, Wenli Li, Ming Zhong, Yuanyuan Zhu, Mengchi Liu, Tieyun Qian

AI总结 本文通过三阶段分析，揭示了LLM在处理异构数据时的格式偏见问题，提出通过数据预处理、推理干预和平衡训练语料库来减少偏见的方法。

Comments Accepted by AAAI 2026, camera ready version

Interpreting Fedspeak with Confidence: A LLM-Based Uncertainty-Aware Framework Guided by Monetary Policy Transmission Paths

以信心解读Fedspeak：一种基于大语言模型的不确定性感知框架，由货币政策传导路径引导

Rui Yao, Qi Chai, Jinhai Yao, Siyuan Li, Junhao Chen, Qi Zhang, Hao Wang

AI总结 本文提出基于大语言模型的不确定性感知框架，用于解读Fedspeak并分类其货币政策立场，通过动态不确定性解码模块提升模型可靠性与准确性。

Comments Accepted by AAAI 2026 Oral

A Differential Perspective on Distributional Reinforcement Learning

分布强化学习的微分视角

Juan Sebastian Rojas, Chi-Guhn Lee

AI总结 本文从微分视角扩展分布强化学习至平均奖励设置，提出首个能有效学习长期每步奖励分布和微分回报分布的算法，并展示其在性能和信息捕捉方面的优势。

Comments In AAAI Conference on Artificial Intelligence 2026

Global Compression Commander: Plug-and-Play Inference Acceleration for High-Resolution Large Vision-Language Models

全局压缩指挥者：面向高分辨率大视觉-语言模型的即插即用推理加速

Xuyang Liu, Ziming Wang, Junjie Chen, Yuhang Han, Yingyao Wang, Jiale Yuan, Jun Song, Siteng Huang, Honggang Chen

机构 * Alibaba（阿里巴巴） 纠错

AI总结 本文提出GlobalCom²框架，通过全局缩略图引导局部裁剪压缩，实现高分辨率LVLMs的高效推理加速。

Comments Accepted by AAAI 2026. Code is available at \url{https://github.com/xuyang-liu16/GlobalCom2}