Distributed lifelong reinforcement learning with sub-linear regret | IEEE Conference Publication | IEEE Xplore