Learning to Play Imperfect-Information Games by Imitating an Oracle Planner
This work provides a more data-efficient method for training agents in complex imperfect-information games, which is significant for researchers and developers working on AI for strategic games.
This paper addresses learning to play multiplayer imperfect-information games with simultaneous moves and large state-action spaces by distilling knowledge from an oracle planner. The oracle, with full state access, uses a fixed-depth tree search and decoupled Thompson sampling for action selection, enabling the follower agent to learn effective strategies in games like Clash Royale and Pommerman with only a few hundred battles.
We consider learning to play multiplayer imperfect-information games with simultaneous moves and large state-action spaces. Previous attempts to tackle such challenging games have largely focused on model-free learning methods, often requiring hundreds of years of experience to produce competitive agents. Our approach is based on model-based planning. We tackle the problem of partial observability by first building an (oracle) planner that has access to the full state of the environment and then distilling the knowledge of the oracle to a (follower) agent which is trained to play the imperfect-information game by imitating the oracle's choices. We experimentally show that planning with naive Monte Carlo tree search does not perform very well in large combinatorial action spaces. We therefore propose planning with a fixed-depth tree search and decoupled Thompson sampling for action selection. We show that the planner is able to discover efficient playing strategies in the games of Clash Royale and Pommerman and the follower policy successfully learns to implement them by training on a few hundred battles.