Attacking cooperative multi-agent reinforcement learning by adversarial minority influence

Simin Li, Jun Guo, Jingqiao Xiu, Yuwei Zheng, Pu Feng, Xin Yu, Jiakai Wang, Aishan Liu, Yaodong Yang, Bo An, Wenjun Wu, Xianglong Liu

Abstract: Highlights•We develop AMI, a strong and practical attack against c-MARL.•Our method deceives victim policy deviation into suboptimal cooperation.•AMI successfully attacks real-world robot swarms and 12 simulation tasks.