Manipulating the Distributions of Experience used for Self-Play Learning in Expert Iteration | IEEE Conference Publication | IEEE Xplore