Opponent Modeling in Multiplayer Imperfect-Information Games
This work addresses the challenge of improving agent performance in strategic multi-agent environments where standard equilibrium-based strategies ignore exploitable opponent behaviors, offering a domain-specific advancement for game AI.
The paper tackles the problem of designing agents for multiplayer imperfect-information games by developing an opponent modeling approach that leverages observations of opponents' play, and it shows that this algorithm significantly outperforms exact Nash equilibrium strategies and other agents in three-player Kuhn poker.
In many real-world settings agents engage in strategic interactions with multiple opposing agents who can employ a wide variety of strategies. The standard approach for designing agents for such settings is to compute or approximate a relevant game-theoretic solution concept such as Nash equilibrium and then follow the prescribed strategy. However, such a strategy ignores any observations of opponents' play, which may indicate shortcomings that can be exploited. We present an approach for opponent modeling in multiplayer imperfect-information games where we collect observations of opponents' play through repeated interactions. We run experiments against a wide variety of real opponents and exact Nash equilibrium strategies in three-player Kuhn poker and show that our algorithm significantly outperforms all of the agents, including the exact Nash equilibrium strategies.