Generating and Adapting to Diverse Ad-Hoc Cooperation Agents in Hanabi
This addresses the challenge of creating diverse test partners for AI agents in cooperative games, which is incremental as it builds on existing methods for behavioral diversity.
The paper tackles the problem of evaluating agents in ad-hoc cooperation settings like Hanabi by generating diverse populations using Quality Diversity algorithms, specifically MAP-Elites, and shows that a meta-strategy for adaptation can outperform generalist strategies when behavioral niches are correctly inferred, though practical interference issues remain.
Hanabi is a cooperative game that brings the problem of modeling other players to the forefront. In this game, coordinated groups of players can leverage pre-established conventions to great effect, but playing in an ad-hoc setting requires agents to adapt to its partner's strategies with no previous coordination. Evaluating an agent in this setting requires a diverse population of potential partners, but so far, the behavioral diversity of agents has not been considered in a systematic way. This paper proposes Quality Diversity algorithms as a promising class of algorithms to generate diverse populations for this purpose, and generates a population of diverse Hanabi agents using MAP-Elites. We also postulate that agents can benefit from a diverse population during training and implement a simple "meta-strategy" for adapting to an agent's perceived behavioral niche. We show this meta-strategy can work better than generalist strategies even outside the population it was trained with if its partner's behavioral niche can be correctly inferred, but in practice a partner's behavior depends and interferes with the meta-agent's own behavior, suggesting an avenue for future research in characterizing another agent's behavior during gameplay.