AAAI

Reference Recommendation based Membership Inference Attack against Hybrid-based Recommender Systems

基于混合推荐系统成员推断攻击的参考推荐

Xiaoxiao Chi, Xuyun Zhang, Yan Wang, Hongsheng Hu, Wanchun Dou

AI总结 本文提出了一种基于指标的成员推断攻击方法，用于检测混合推荐系统中用户数据的成员资格，通过个性化特征和参考推荐来提高攻击效果。

Comments This paper has been accepted by AAAI 2026

ASSIST-3D: Adapted Scene Synthesis for Class-Agnostic 3D Instance Segmentation

ASSIST-3D: 为无类别3D实例分割适应的场景合成

Shengchao Zhou, Jiehong Lin, Jiahui Liu, Shizhen Zhao, Chirui Chang, Xiaojuan Qi

AI总结 ASSIST-3D通过异构对象选择、场景布局生成和点云构建，提升无类别3D实例分割的泛化能力。

Comments Accepted by AAAI 2026

LongT2IBench: A Benchmark for Evaluating Long Text-to-Image Generation with Graph-structured Annotations

LongT2IBench: 一个用于评估长文本到图像生成的基准，具有图结构注释

Zhichao Yang, Tianjiao Gu, Jianjie Wang, Feiyu Lin, Xiangfei Sheng, Pengfei Chen, Leida Li

AI总结 LongT2IBench通过图结构注释和LongT2IExpert提出，用于评估长文本到图像生成的对齐和解释能力。

Comments The paper has been accepted by AAAI 2026

Enabling Delayed-Full Charging Through Transformer-Based Real-Time-to-Departure Modeling for EV Battery Longevity

通过基于Transformer的实时到出发建模实现延迟满充以延长电动汽车电池寿命

Yonggeon Lee, Jibin Hwang, Alfred Malengo Kondoro, Juhyun Song, Youngtae Noh

AI总结 本文提出基于Transformer的实时到出发建模方法，通过预测用户出发时间实现延迟满充，延长电动汽车电池寿命，提升可持续交通系统的实用性。

Comments 16 pages, 9 figures, AAAI'26 (accepted)

PathMind: A Retrieve-Prioritize-Reason Framework for Knowledge Graph Reasoning with Large Language Models

PathMind: 一种基于大语言模型的知识图谱推理检索-优先-推理框架

Yu Liu, Xixun Lin, Yanmin Shang, Yangxi Li, Shi Wang, Yanan Cao

AI总结 PathMind通过检索-优先-推理框架，提升大语言模型在知识图谱推理中的准确性和可解释性，尤其在复杂推理任务中表现优异。

Comments AAAI 2026, Long Paper, Oral

GloTok: Global Perspective Tokenizer for Image Reconstruction and Generation

GloTok: 用于图像重建和生成的全局视角分词器

Xuan Zhao, Zhongyu Zhang, Yuge Huang, Yuxi Mi, Guodong Mu, Shouhong Ding, Jun Wang, Rizen Guo, Shuigeng Zhou

机构 * Tencent Youtu Lab（腾讯优图实验室） 纠错

AI总结 GloTok通过全局视角分词器利用全局关系信息，实现更均匀的语义分布，提升图像生成质量。

Comments Accepted at AAAI'26

New Synthetic Goldmine: Hand Joint Angle-Driven EMG Data Generation Framework for Micro-Gesture Recognition

新的合成金矿：基于手关节角度驱动的EMG数据生成框架用于微手势识别

Nana Wang, Gen Li, Pengfei Ren, Hao Su, Suli Wang

AI总结 本文提出SeqEMG-GAN框架，通过手关节角度驱动生成高保真EMG数据，提升微手势识别性能和数据集多样性。

Comments Accepted at AAAI 2026

MACS: Multi-source Audio-to-image Generation with Contextual Significance and Semantic Alignment

MACS：基于上下文意义和语义对齐的多源音频到图像生成

Hao Zhou, Xiaobao Guo, Yuzhe Zhu, Adams Wai-Kin Kong

AI总结 MACS通过分离多源音频并利用语义对齐提升音频到图像生成的质量和表现。

Comments Accepted at AAAI 2026. Code available at https://github.com/alxzzhou/MACS