An Implementation of Monte Carlo Tree Search Algorithm: Comparison with Random Samples

MCTS is best suit for games like chess, poker, etc. It works well in terms of performance in many of interesting gambling-game arenas, frequently crossing the human expertise. MCTS is extensively used in domains where extensive planning, scheduling, optimization is required. The major features of MCTS is its search-based methodology to handle games of unknown terrain and imperfect paths. In addition to this, MCTS can be parallelized, scalable and thus fits for distributed computing multi-core platforms. In this paper we use MCTS for Tic-Tok-Toe board game and see how the Randomized playouts and MCTS agent work with minimal Computational Resources and less combination of playouts.