Monte Carlo Tree Search (MCTS) is a heuristic search algorithm that can make game play decisions. It does this by attempting to determine the most promising move from those available. What’s interesting is that the algorithm just needs to know the game rules and winning conditions, it doesn’t need domain knowledge. Another advantage is that it can be interrupted or constrained by a time limit and will produce the most promising move found so far.
I built a Tic Tac Toe player a couple of years ago and while upgrading Unity recently I thought I’d try building it for the web so I built a WebGL version.
Continue reading “Monte Carlo Tree Search demo”