I've been learning how the Monte Carlo Tree Search works and implemented it to play a game of Connect 4. I think I've got the algorithm correct as it plays the game quite well (it can usually beat me extremely quickly), but I was curious to see how it played against the version at https://connect4.gamesolver.org/en. My AI puts up a reasonable fight but always loses eventually. In the blog for that site it explains how it's using Minimax + alpha beta pruning and a load of extra optimisations I don't understand (I'm still new to this).

So my question is should an AI based on MCTS be able to play as well as an optimised Minimax algorithm and if not, what kinds of optimisations and improvements should I be looking at making to get it to that level?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.