This tutorial walks through a synchronous single-thread single-GPU (read malnourished) game-agnostic implementation of the recent AlphaGo Zero paper by DeepMind. It's a beautiful piece of work that trains an agent for the game of Go through pure self-play without any human knowledge except the rules of the game. The methods are fairly simple compared to previous papers by DeepMind, and AlphaGo Zero ends up beating AlphaGo (trained using data from expert games and beat the best human Go players) convincingly. Recently, DeepMind published a preprint of Alpha Zero on arXiv that extends AlphaGo Zero methods to Chess and Shogi.

The aim of this post is to distil out the key ideas from the AlphaGo Zero paper and understand them concretely through code. It assumes basic familiarity with machine learning and reinforcement learning concepts, and should be accessible if you understand neural network basics and Monte Carlo Tree Search. Before starting out (or after finishing this tutorial), I would recommend reading the original paper. It's well-written, very readable and has beautiful illustrations! AlphaGo Zero is trained by self-play reinforcement learning. It combines a neural network and Monte Carlo Tree Search in an elegant policy iteration framework to achieve stable learning. But that's just words- let's dive into the details straightaway.

[..]

... but have no clue about machine learning at all and this is very welcome.