Stanford CS330: Multi-Task and Meta-Learning, 2019 | Lecture 6 - Reinforcement Learning Primer

For more information about Stanford’s Artificial Intelligence professional and graduate programs, visit: https://stanford.io/ai Assistant Professor Chelsea Finn, Stanford University http://cs330.stanford.edu/ 0:00 Introduction 0:46 Logistics 2:31 Why Reinforcement Learning? 3:37 The Plan 6:16 Terminology & notation 8:36 Imitation Learning 10:01 Reward functions 10:57 The goal of reinforcement learning 19:15 What is a reinforcement learning task? 21:01 The goal of multi-task reinforcement learning 23:31 The anatomy of a reinforcement learning algorithm 25:48 Evaluating the objective 26:43 Direct policy differentiation 32:02 Evaluating the policy gradient 33:16 Comparison to maximum likelihood 35:54 Example: MAML + policy gradient 37:25 Example: Black-box meta-learning + policy gradient 45:26 Policy Gradients 49:16 Value-Based RL: Definitions 52:14 Fitted Q-iteration Algorithm 56:13 Multi-Task RL Algorithms 58:00 An example