On Practical Reinforcement Learning: Provable Robustness, Scalability, and Statistical Efficiency
thesis
posted on 2023-07-28, 03:31authored byTang Thanh Nguyen
This thesis presented a novel and unifying approach to close the gap of several current reinforcement learning (RL) methods in practical considerations, including robust RL, distributional RL, and offline RL. It provided practical algorithms with strong theoretical guarantees, with a chief focus on provable robustness, scalability and statistical efficiency.