Mastering the game of Go without human knowledge

Abstract

This paper introduces and evaluates the idea described in “Mastering the game of Go without human knowledge”, and reports empirical results that helped shape subsequent work in reinforcement-learning, alphago.