A very simple Game to test some AI principles !
This is a very simple Tic-Tac-Toe against an AI which learn automatically. You play cross, AI plays circle. To restart a game, you nee to press ENTER. To save what AI learnt, press SPACE.
The approach is as follow : the AI registers possible state of the game, and associates with each of them couples of datas : an available position to play and the probability to play this position at this game state. If it discovers a new state, it simply initializes uniform probabilities. At the end of the game, the AI modify its probabilities regarding where did it plays and which position it choosed.
The probabilities change policy isn't always easy to determine, and that is why this repository exist ! To test some policies.