IJCAI

Tackling Multimodal Learning Challenges with Mixture-of-Expert: A Survey

应对多模态学习挑战的混合专家方法：综述

Liangwei Nathan Zheng, Wei Emma Zhang, Olaf Maennel, Lin Yue, Weitong Chen

机构 * Adelaide University（阿德莱德大学） 纠错

AI总结 本文综述了混合专家（MoE）如何通过高效扩展、表示学习和自适应适配解决多模态学习中的可扩展性、异质性和数据不完美等核心挑战。

Comments This survey paper has just been accepted by IJCAI 2026. Results were released by 30 April 2026. As I could not find a particular place to drop the acceptance email. I have upload the acceptance email alongside the LaTeX files of the paper, named as Acceptance_email.pdf

Can Quantum Federated Learning Withstand Circuit-Level Backdoors?

量子联邦学习能否抵御电路级后门攻击？

Aakar Mathur, Mohammed Ruknuddin, Ashish Gupta

机构 * BITS Pilani Dubai Campus（比斯汉尼迪拜校区） 纠错

AI总结 提出电路级后门威胁模型（CULT），通过量子感知机制（Grover、Pauli、Bit-flip、Sign-flip）实现四种隐蔽攻击，理论证明攻击的隐蔽性，实验表明单个恶意客户端即可导致FedAvg精度严重下降，现有防御无法消除最坏情况。

Comments Accepted to IJCAI-ECAI 2026

Inversely Learning Transferable Rewards via Abstracted States

通过抽象状态逆向学习可迁移奖励

Yikang Gui, Prashant Doshi

机构 * THINC Lab, School of Computing University of Georgia（THINC实验室，计算学院，佐治亚大学） ； School of Computing and Institute for AI University of Georgia（计算学院和人工智能研究所，佐治亚大学） 纠错

AI总结 提出一种通过行为轨迹逆向学习抽象奖励函数的方法，并在未见过的领域实例中验证其可迁移性。

Comments Accepted at IJCAI 2026