Adversarial attacks in consensus-based multi-agent reinforcement learning
Adversarial attacks in consensus-based multi-agent reinforcement learning
复制标题
DOI：
10.23919/acc50511.2021.9483080
复制
发表时间：
2021-03
期刊：
影响因子：
--
通讯作者：
Martin Figura;K. Kosaraju;V. Gupta
中科院分区：
文献类型：
--
作者：
Martin Figura;K. Kosaraju;V. Gupta
近年来，许多合作分布式多智能体强化学习（MARL）算法已被提出在文献中。在这项工作中，我们研究了对抗性攻击对采用基于共识的MARL算法的网络的影响。我们证明了一个对抗性代理可以说服网络中的所有其他代理实施优化其期望目标的策略。从这个意义上说，标准的基于共识的MARL算法是脆弱的攻击。
Recently, many cooperative distributed multiagent reinforcement learning (MARL) algorithms have been proposed in the literature. In this work, we study the effect of adversarial attacks on a network that employs a consensus-based MARL algorithm. We show that an adversarial agent can persuade all the other agents in the network to implement policies that optimize an objective that it desires. In this sense, the standard consensus-based MARL algorithms are fragile to attacks.