Learning to Play Tic Tac Toe

To show how adaptable neural networks can be, we will attempt to use a neural network to learn the optimal moves of Tic Tac Toe. We will approach this knowing that Tic Tac Toe is a deterministic game and that the optimal moves are already known.

Getting ready

To train our model, we will have a list of board positions followed by the best optimal response for a number of different boards. We can reduce the amount of boards to train on by considering only board positions that are different with respect to symmetries. The non-identity transformations of a Tic Tac Toe board are a rotation (either direction) by 90 degrees, 180 degrees, 270 degrees, a horizontal reflection, and a vertical reflection. Given this idea, we will use a shortlist of boards with the optimal move, apply two random transformations, and feed that into our neural network to learn.

Note

Since tic-tac-toe is a deterministic game, it is worth noting that whoever goes first should either win or draw...