Exploiting Game Decompositions in Monte Carlo Tree Search

Aline Hufschmitt (1) , Jean-Noël Vittaut (2) , Nicolas Jouandeau (3)

1 CREC - Centre de Recherche des Ecoles de St-Cyr Coëtquidan
2 LFI - Learning, Fuzzy and Intelligent systems
3 UP8 - Université Paris 8 Vincennes-Saint-Denis

In this paper, we propose a variation of the MCTS framework to perform a search in several trees to exploit game decompositions. Our Multiple Tree MCTS (MT-MCTS) approach builds simultaneously multiple MCTS trees corresponding to the different sub-games and allows , like MCTS algorithms, to evaluate moves while playing. We apply MT-MCTS on decomposed games in the General Game Playing framework. We present encouraging results on single player games showing that this approach is promising and opens new avenues for further research in the domain of decomposition exploitation. Complex compound games are solved from 2 times faster (Incredible) up to 25 times faster (Nonogram).

Mots clés: Monte Carlo Tree Search; General Game Playing; Decomposition

Domaines: Intelligence artificielle [cs.AI]

Aline Hufschmitt, Jean-Noël Vittaut, Nicolas Jouandeau. Exploiting Game Decompositions in Monte Carlo Tree Search. 16th Advances in Computer Games Conference, Aug 2019, Macao, China. ⟨hal-02317329v2⟩