Multi-Agent RL

RL with multiple interacting agents that cooperate or compete. Self-play in AlphaGo Zero and multi-agent training in AlphaStar are landmark examples.