Gu, S. (2019). Sample-efficient deep reinforcement learning for continuous control. PhD Thesis, University of Cambridge, Cambridge.