Skip to content

Example TensorFlow experiment - using Machine Learning to build a Noughts And Crosses player, inspired by Google DeepMind's AlphaGo Zero.

License

Notifications You must be signed in to change notification settings

danlester/TensorFlow-NoughtsAndCrosses

Repository files navigation

Noughts and Crosses

Most tutorials for machine learning ultimately end up in a script that outputs a number such as accuracy that might increase from 60% as a baseline to 92% after applying an ML algorithm. The training and evaluation processes are both part of the same script.

It's just not very inspiring - I wanted to build something using ML that I could interact with, and feel how the ML techniques have actually contributed to it.

So I wanted a 'playground' project to experiment with the available techniques.

Inspired by AlphaGo Zero, I decided to build a Noughts and Crosses (Tic Tac Toe) playground, to see where that takes me. Crucially, if I can build a Machine Intelligence robot to play the game, I can get it to play against myself (and itself...)

This tutorial will explain the project and lead you through generating data to train various Machine Learning models, allowing you to play against the Machine Intelligence you've built. It assumes you have already installed TensorFlow on your machine and followed some of the basic 'deep learning' examples. The journey presented is my own experience experimenting with TensorFlow to understand how to use it. Presumes Python knowledge and ability to install packages etc.

You can read the tutorial here.

About

Example TensorFlow experiment - using Machine Learning to build a Noughts And Crosses player, inspired by Google DeepMind's AlphaGo Zero.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages