Automatic Grouping for Efficient Cooperative Multi-Agent Reinforcement Learning

Multi-Agent Reinforcement Learning (GoMARL), which learns automatic grouping without domain knowledge for efficient cooperation.