Toward Generalization of Automated Temporal Abstraction to Partially Observable Reinforcement Learning | IEEE Journals & Magazine | IEEE Xplore