Lecture 6: Exploring Model-Based and Model-Free Reinforcement Learning Algorithms

In this ECE 8851: Reinforcement Learning lecture, we discuss model-based and model-free reinforcement learning algorithms. We start off by discussing the asynchronous value and policy iteration algorithms, which are model-based reinforcement learning algorithms. We then move on to the model-free reinforcement learning algorithms, where we cover the Q-learning, enhanced policy iteration, and SARSA algorithms.

Throughout the lecture, we focus on the key differences among these algorithms, including on-policy RL, tapering step size, update function, dampening, and EPI. We examine the advantages and disadvantages of each algorithm and discuss how they can be used to solve complex reinforcement learning problems.