Week 3: Reinforcement Learning

This week, you will learn about reinforcement learning, and build a deep Q-learning neural network in order to land a virtual lunar lander on Mars!

Learning Objectives

• Understand key terms such as return, state, action, and policy as it applies to reinforcement learning

• Understand the Bellman equations

• Understand the state-action value function

• Understand continuous state spaces

• Build a deep Q-learning network

Reinforcement learning introduction

What is Reinforcement Learning? - Video • Duration: 8 min

Mars rover example - Video • Duration: 6 min

The Return in reinforcement learning - Video • Duration: 10 min

Making decisions: Policies in reinforcement learning - Video • Duration: 2 min Review of key concepts - Video • Duration: 5 min

Practice quiz: Reinforcement learning introduction

Reinforcement learning introduction

State-action value function

State-action value function definition - Video • Duration: 10 min

State-action value function example - Video • Duration: 5 min

Bellman Equations - Video • Duration: 12 min

Random (stochastic) environment (Optional) - Video • Duration: 8 min

State-action value function (optional lab) - Lab • Duration: 1 h

Quiz: State-action value function

State-action value function

Continuous state space

Example of continuous state space applications - Video • Duration: 6 min

Lunar lander - Video • Duration: 5 min

Learning the state-value function - Video • Duration: 16 min

Algorithm refinement: Improved neural network architecture - Video • Duration: 3 min

Algorithm refinement: ϵ-greedy policy - Video • Duration: 8 min

Algorithm refinement: Mini-batch and soft updates (optional) - Video • Duration: 11 min

The state of reinforcement learning - Video • Duration: 2 min

Quiz: Continuous state spaces

Continuous state spaces

Practice Lab: Reinforcement learning

Reinforcement Learning

Summary and thank you

Summary and thank you - Video • Duration: 3 min

Acknowledgments

Acknowledgments