Othello mini-game showing how to integrate a neural network into the gameplay loop to drive a boardgame opponent.
We want to code an opponent that can play our boardgame.
Our boardgame can be somewhat complex with specific rules and opponent interactions and difficulty curves.
To solve this problem we leverage a small Neural Network.
It takes as input the board state.
After running inference we get for every position on the board a possible move with its corresponding probability to win as well as who is currently winning.
We use this to feedback into the gameplay loop
