Ian Osband
TalkRL: The Reinforcement Learning Podcast
00:00
Exploring Thompson Sampling Approximation in Reinforcement Learning
The chapter discusses the challenges and importance of approximating Thompson sampling in reinforcement learning, highlighting the significance of behavior suites and introducing the 'deep sea' task to illustrate exploration difficulties in RL. It also introduces the concept of using deep neural networks and bootstrap DQN to efficiently solve complex tasks.
Transcript
Play full episode
