Methodology / Approach

The approach adopted is that of Reinforcement Learning. Two neural networks compete against each other to understand a nuanced board game and learn it's rules and victory strategies all by themselves, with just the knowledge of wins/losses and draws. The following link to my blogpost showcases the same :
https://kashikarparth.github.io/posts/q-learning-powered-gamebot

Technologies Used

Intel Distribution for Python, TensorFlow* Optimizations for the Intel Architecture

Repository

https://github.com/kashikarparth/4x4TicTacToe