Sparse Adversarial Attack in Multi-agent Reinforcement Learning

2022/05/19 by Yizheng Hu, Zhihua Zhang, Hu, Yizheng +1 · 1 citation
Computer Science · Medicine · #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Mosquito-borne diseases and control

paper · pdf · doi:10.48550/arxiv.2205.09362

openalex publication_date 2022/05/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Cooperative multi-agent reinforcement learning (cMARL) has many real applications, but the policy trained by existing cMARL algorithms is not robust enough when deployed. There exist also many methods about adversarial attacks on the RL system, which implies that the RL system can suffer from adversarial attacks, but most of them focused on single agent RL. In this paper, we propose a sparse adversarial attack on cMARL systems. We use (MA)RL with regularization to train the attack policy. Our experiments show that the policy trained by the current cMARL algorithm can obtain poor performance when only one or a few agents in the team (e.g., 1 of 8 or 5 of 25) were attacked at a few timesteps (e.g., attack 3 of total 40 timesteps).

Cited by

• Constrained Black-Box Attacks Against Multi-Agent Reinforcement Learning

Related

• Attacking c-MARL More Effectively: A Data Driven Approach
• Constrained Black-Box Attacks Against Multi-Agent Reinforcement Learning
• Adversarial attacks in consensus-based multi-agent reinforcement\n learning
• Efficient Adversarial Attacks on Online Multi-agent Reinforcement Learning
• Robust multi-agent coordination via evolutionary generation of auxiliary adversarial attackers