Adversarial attacks in consensus-based multi-agent reinforcement learning

Martin Figura; Krishna Chaitanya Kosaraju; Vijay Gupta

Adversarial attacks in consensus-based multi-agent reinforcement learning

Systems and Control 2021-03-15 v1 Artificial Intelligence Machine Learning Systems and Control Machine Learning

Authors: Martin Figura , Krishna Chaitanya Kosaraju , Vijay Gupta

Abstract

Recently, many cooperative distributed multi-agent 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.

Keywords

multi-agent systems and game theory distributed consensus cyber-physical system security

Cite

@article{arxiv.2103.06967,
  title  = {Adversarial attacks in consensus-based multi-agent reinforcement learning},
  author = {Martin Figura and Krishna Chaitanya Kosaraju and Vijay Gupta},
  journal= {arXiv preprint arXiv:2103.06967},
  year   = {2021}
}

Related papers

View all related →