Understanding AlphaGo -A Reinforcement Learning Perspective
Abstract
Recently, AlphaGo has gained worldwide attention for its success in achieving super-human performance in the ancient Go game. However, it's still mysterious for most laymen, who may overestimate today's AI research. The aim of this paper is to dissect all the algorithms used in building AlphaGo from a reinforcement learning perspective, which cannot be found in its seminal paper.