Transfer reinforcement learning for fault-tolerant control by re-using optimal policies | IEEE Conference Publication | IEEE Xplore