A Hybrid Multi-Task Learning Approach for Optimizing Deep Reinforcement Learning Agents | IEEE Journals & Magazine | IEEE Xplore