A sampled fictitious play based learning algorithm for infinite horizon Markov Decision Processes | IEEE Conference Publication | IEEE Xplore