Lecture 23: Reinforcement Learning I

Content locked

The catalog is public. Reading this content requires approved access.

Introduction to Reinforcement Learning

Real-World Applications

Markov Decision Process (MDP)

Components of an MDP

The Markov Assumption

Example: Grid World

Reinforcement Learning Challenges

Credit Assignment Problem

Exploration vs. Exploitation

Defining the Optimal Policy

Value Function

Discounting Rewards

From Value to Policy

Bellman Equation

Value Iteration

Q-Learning

The Q-Function

Q-Learning Update Rule

Epsilon-Greedy Exploration

Q-Learning Algorithm

Practice Problems

Key Takeaways