An improved approach to reinforcement learning in Computer Go | IEEE Conference Publication | IEEE Xplore