<ul><li>Reinforcement learning in partially observable environments is challenging, especially in multi-agent settings.</li><li>The authors propose using learned beliefs on the underlying system state to overcome these challenges.</li><li>Belief states are pre-trained in a self-supervised fashion and used in a state-based reinforcement learning algorithm.</li><li>The proposed method simplifies learning tasks, improves convergence speed, and final performance.</li></ul>

Belief States for Cooperative Multi-Agent Reinforcement Learning under Partial Observability

Discover more