A Joint Imitation-Reinforcement Learning Framework for Reduced Baseline Regret | IEEE Conference Publication | IEEE Xplore