AAAI

Improving the Convergence Rate of Ray Search Optimization for Query-Efficient Hard-Label Attacks

提升查询效率的射线搜索优化在硬标签攻击中的收敛速度

Xinjie Xu, Shuyu Cheng, Dongwei Xu, Qi Xuan, Chen Ma

AI总结 本文提出ARS-OPT和PARS-OPT算法，通过优化射线搜索方向提升硬标签攻击的收敛速度和查询效率。

Comments Published at AAAI 2026 (Oral). This version corresponds to the conference proceedings; v2 will include the appendix

USE: A Unified Model for Universal Sound Separation and Extraction

USE: 一种统一的通用声音分离与提取模型

Hongyu Wang, Chenda Li, Xin Zhou, Shuai Wang, Yanmin Qian

AI总结 本文提出USE模型，通过统一框架结合声音分离与目标声音提取，提升两者性能，实验显示在SS和TSE任务中分别取得1.4 dB SDR提升和86%准确率。

Comments Accepted as an oral presentation by AAAI 2026

FreeInpaint: Tuning-free Prompt Alignment and Visual Rationality Enhancement in Image Inpainting

FreeInpaint: 图像修复中无调优的提示对齐与视觉合理性增强

Chao Gong, Dong Li, Yingwei Pan, Jingjing Chen, Ting Yao, Tao Mei

AI总结 FreeInpaint通过无调优的扩散潜在特征优化方法，提升图像修复的提示对齐和视觉合理性。

Comments Accepted by AAAI 2026

Towards Better Search with Domain-Aware Text Embeddings for C2C Marketplaces

面向C2C市场向更好搜索的领域感知文本嵌入

Andre Rusli, Miao Cao, Shoma Ishimoto, Sho Akiyama, Max Frenzel

AI总结 本文提出领域感知的文本嵌入方法，通过优化C2C市场places的搜索质量，提升相关性和效率，为更丰富的LLM时代搜索体验奠定基础。

Comments 5 pages, AAAI 2026 Workshop on New Frontiers in Information Retrieval

Self-supervised Multiplex Consensus Mamba for General Image Fusion

自监督多乘共识Mamba用于通用图像融合

Yingying Wang, Rongjin Zhuang, Hui Zheng, Xuanhua He, Ke Cao, Xiaotong Tu, Xinghao Ding

AI总结 SMC-Mamba通过自监督多乘共识机制提升通用图像融合性能，有效整合多模态信息并优化下游任务表现。

Comments Accepted by AAAI 2026, 9 pages, 4 figures

Soft Filtering: Guiding Zero-shot Composed Image Retrieval with Prescriptive and Proscriptive Constraints

软过滤：通过规劝性与禁止性约束指导零样本复合图像检索

Youjin Jung, Seongwoo Cho, Hyun-seok Min, Sungchul Choi

AI总结 本文提出SoFT方法，通过规劝性和禁止性约束提升零样本复合图像检索的准确性与鲁棒性。

Comments Accepted to AAAI 2026 Workshop on New Frontiers in Information Retrieval

Learning to Compress: Unlocking the Potential of Large Language Models for Text Representation

学习压缩：解锁大型语言模型在文本表示中的潜力

Yeqin Zhang, Yizheng Zhao, Chen Hu, Binxing Jiao, Daxin Jiang, Ruihang Miao, Cam-Tu Nguyen

AI总结 本文提出通过上下文压缩作为预训练任务来提升大型语言模型的文本表示能力，实验表明其在多种任务中优于传统方法。

Comments Accepted by AAAI'26

O3SLM: Open Weight, Open Data, and Open Vocabulary Sketch-Language Model

O3SLM：开放权重、开放数据和开放词汇草图-语言模型

Rishi Gupta, Mukilan Karuppasamy, Shyam Marjit, Aditay Tripathi, Anirban Chakraborty

机构 * IISc（印度理工学院） 纠错

AI总结 O3SLM通过构建大规模图像-草图-指令三元组数据集和训练模型，实现了对草图理解和推理的突破性进展。

Comments Accepted to AAAI 2026

Learning Fair Representations with Kolmogorov-Arnold Networks

利用柯尔莫戈罗夫-阿诺尔德网络学习公平表示

Amisha Priyadarshini, Sergio Gago-Masague

AI总结 本文提出利用柯尔莫戈罗夫-阿诺尔德网络在公平对抗学习框架中实现公平表示，通过理论分析和实证研究提升公平性与预测准确性的平衡。

Comments This article has been accepted at AAAI-26 (The 40th Annual AAAI Conference on Artificial Intelligence)

RGMP: Recurrent Geometric-prior Multimodal Policy for Generalizable Humanoid Robot Manipulation

RGMP: 基于几何先验的多模态策略用于通用人形机器人操作

Xuetao Li, Wenke Huang, Nengyuan Pan, Kaiyan Zhao, Songhua Yang, Yiming Wang, Mengde Li, Mang Ye, Jifeng Xuan, Miao Li

AI总结 RGMP通过结合几何-语义推理与递归高斯适应，实现了高效的人形机器人多模态操作控制。

Journal ref Proceedings of the AAAI conference on artificial intelligence, 2026

Rethinking Direct Preference Optimization in Diffusion Models

重新思考扩散模型中的直接偏好优化

Junyong Kang, Seohyun Lim, Kyungjune Baek, Hyunjung Shim

AI总结 本文提出了一种改进扩散模型偏好优化的方法，通过稳定参考模型更新和时间步感知训练策略，提升模型在人类偏好评估中的性能。

Comments Accepted by SPIGM@NeurIPS 2025 and AAAI-26 (Oral)

AI Supply Chains: An Emerging Ecosystem of AI Actors, Products, and Services

AI供应链：人工智能行为者、产品和服务的新兴生态系统

Aspen Hopkins, Sarah H. Cen, Andrew Ilyas, Isabella Struckman, Luis Videgaray, Aleksander Mądry

AI总结 本文研究了人工智能供应链的形成与影响，通过案例分析揭示了供应链中信息传递的不完善及上游设计对下游的连锁影响。

Comments 27 pages, 8 figures

Journal ref Proc. AAAI/ACM Conf. AI Ethics Soc. (AIES), 8(2), 1266-1277 (2025)