Elastic Monte Carlo Tree Search with State Abstraction for Strategy Game Playing

TL;DR

This paper introduces Elastic MCTS, a dynamic state abstraction algorithm for strategy game AI, which improves search efficiency and performance by adaptively clustering game states during Monte Carlo Tree Search.

Contribution

The paper presents Elastic MCTS, a novel algorithm that dynamically clusters game states during search, reducing complexity without domain-specific knowledge.

Findings

Elastic MCTS outperforms baseline MCTS in strategy games.

The method reduces tree size by a factor of 10.

Elastic MCTS scales well with game complexity.

Abstract

Strategy video games challenge AI agents with their combinatorial search space caused by complex game elements. State abstraction is a popular technique that reduces the state space complexity. However, current state abstraction methods for games depend on domain knowledge, making their application to new games expensive. State abstraction methods that require no domain knowledge are studied extensively in the planning domain. However, no evidence shows they scale well with the complexity of strategy games. In this paper, we propose Elastic MCTS, an algorithm that uses state abstraction to play strategy games. In Elastic MCTS, the nodes of the tree are clustered dynamically, first grouped together progressively by state abstraction, and then separated when an iteration threshold is reached. The elastic changes benefit from efficient searching brought by state abstraction but avoid the negative influence of using state abstraction for the whole search. To evaluate our method, we make use of the general strategy games platform Stratega to generate scenarios of varying complexity. Results show that Elastic MCTS outperforms MCTS baselines with a large margin, while reducing the tree size by a factor of $10$. Code can be found at: https://github.com/egg-west/Stratega