(Presented by Vladlen Koltun in CS 294-131.)
This paper replaces the traditional reward formulation of reinforcement learning problems into a reward stream.
See: https://danieltakeshi.github.io/2017/10/10/learning-to-act-by-predicting-the-future/
(Presented by Vladlen Koltun in CS 294-131.)
This paper replaces the traditional reward formulation of reinforcement learning problems into a reward stream.
See: https://danieltakeshi.github.io/2017/10/10/learning-to-act-by-predicting-the-future/