Learning Deep Neural Network Policies with Continuous Memory States

Marvin Zhang, Zoe McCarthy, Chelsea Finn +2

Policy learning for partially observed control tasks requires policies that can remember salient information from past observations. In this paper, we present a method for learning…