Reinforcement Learning — Finite Markov Decision Processes and Dynamic Programming | Computer Science Online Tutorial
TUTORIAL 099 / 100 Reinforcement Learning A self-paced online tutorial session on finite markov decision processes and dynamic programming, part of the Reinforcement Learning track. Work through the topics below, then reach out any time if you'd like a live walkthrough. Estimated session length: 9 hours Format: Online, self-paced What this tutorial covers 01 The Agent-Environment Interface 02 Goals and Rewards 03 Finite Markov Decision Process 04 The Agent-Environment Interface 05 Goals and Rewards 06 Returns and Episodes 07 Unified Notation for Episodic and Continuing Tasks 08 Policies and Value Functions 09 Optimal Policies and Optimal Value Functions 10 Optimality and Approximation 11 Dynamic Programming 12 Policy Evaluation (Prediction) 13 Policy Improvement 14 Policy Iteration 15 Value Iteration 16 Generalized Policy Iteration 17 Efficiency of Dynamic Programming 18 Asynchronous Dynamic Programming T3 Policies and value functions for Gridworld ex...