PROJECT TITLE :

Information Set Monte Carlo Tree Search

ABSTRACT:

Monte Carlo tree search (MCTS) is an AI technique that has been successfully applied to many deterministic games of good info. This paper investigates the applying of MCTS methods to games with hidden data and uncertainty. In particular, three new information set MCTS (ISMCTS) algorithms are presented that handle totally different sources of hidden information and uncertainty in games. Rather than looking minimax trees of game states, the ISMCTS algorithms search trees of knowledge sets, additional directly analyzing the true structure of the sport. These algorithms are tested in three domains with completely different characteristics, and it is demonstrated that our new algorithms outperform existing approaches to handling hidden info and uncertainty in games.