The k-server problem: a reinforcement learning approach | IEEE Conference Publication | IEEE Xplore