Cooperative Actor-Critic via TD Error Aggregation

Martin Figura, Yixuan Lin, Ji Liu +1

In decentralized cooperative multi-agent reinforcement learning, agents can aggregate information from one another to learn policies that maximize a team-average objective function…