AlphaGo and General Learning Methods

AlphaGo combined learned value and policy networks with tree search to play Go at a high level. Its successors reduced dependence on human game data and applied related reinforcement-learning and search ideas across multiple games and optimization problems.