Tic-Tac-Toe AI

Trained MLP — try to beat it!

Choose who goes first
The AI was trained using minimax-optimal data on a Multi-Layer Perceptron. Board is normalised so the model always sees itself as +1.