AlphaDraughts Zero

AlphaGo Zero is a historical breakthrough in Game AI, not only because that it beated all the previous versions and all human experts, but also because that it is learnt by self-play without any human knowledge.

In this project, we investigate at first the algorithm flow of AlphaGo Zero. A readily compmrehensible interpretation is provided. We then apply the algorithm in English Draughts.

Requirement

• Python 3, version 3.4 or later is preferred
• PyTorch 1.0
• CUDA, CuDNN (If GPU is available)

For pip users, run pip install -r requirements.txt to install dependencies.

Source Code

Code / Report / Slide