American Go E-Journal

Review: Deep Learning and the Game of Go

Thursday March 14, 2019

Deep Learning and the Game of Go  by Max Pumperla and Kevin Ferguson, teaches all the aspects of deep learning that went 2019.03.13_DeepLearning-Pumperla-DL-HIinto Alpha Go. It is complete with code in Python3, with all code based upon building a go-playing program starting with code for the board structures/objects. It does not teach Python and does not assume you know anything about learning algorithms. I am only in the 4th chapter and so far I find the text very readable and the explanations clear. By the end of the book you can have a complete Go bot for playing or experimenting. One very nice thing that Manning does is that all the code is available for downloading, and buying the paper book allows you to signup for electronic versions in a variety of formats, so you can copy/paste the examples if you don’t want to type them and you can search the text electronically rather than paging through the book. About the only thing not covered in the book is parallelization for multiple GPU/CPUs.
- David Doshay