Multitask Neuroevolution for Reinforcement Learning With Long and Short Episodes | IEEE Journals & Magazine | IEEE Xplore