A Communication-Efficient Multi-Agent Actor-Critic Algorithm for Distributed Reinforcement Learning
This addresses communication efficiency in multi-agent systems, which is incremental as it builds on existing actor-critic methods.
The paper tackles the problem of distributed reinforcement learning with multiple agents communicating only locally, proposing a randomized actor-critic algorithm that allows each agent to transmit only two scalar variables per time step, achieving a solution for strongly connected directed graphs.
This paper considers a distributed reinforcement learning problem in which a network of multiple agents aim to cooperatively maximize the globally averaged return through communication with only local neighbors. A randomized communication-efficient multi-agent actor-critic algorithm is proposed for possibly unidirectional communication relationships depicted by a directed graph. It is shown that the algorithm can solve the problem for strongly connected graphs by allowing each agent to transmit only two scalar-valued variables at one time.