Machine Learning implementation for TicTacToe. Creates two AIs which plays 10 million games against each other. Increases weights of winning moves, decreases weight of losing moves in the path of that game. Weights determine probability that the move will be played in that specific board state. If tie, no change. Uses ternary to store unique board states by storing X as 1, O as 2, and an empty space as 0.
Plans to try a different approach. Will try to add a direct heuristics approach storing only 6 weights as opposed to storing every possible board state.