Policy Correction and State-Conditioned Action Evaluation for Few-Shot Lifelong Deep Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore