Distributional Reinforcement Learning
In this chapter, we will learn about distributional reinforcement learning. We will begin the chapter by understanding what exactly distributional reinforcement learning is and why it is useful. Next, we will learn about one of the most popular distributional reinforcement learning algorithms called categorical DQN. We will understand what a categorical DQN is and how it differs from the DQN we learned in Chapter 9, Deep Q Networks and Its Variants, and then we will explore the categorical DQN algorithm in detail.
Following this, we will learn another interesting algorithm called Quantile Regression DQN (QR-DQN). We will understand what a QR-DQN is and how it differs from a categorical DQN, and then we will explore the QR-DQN algorithm in detail.
At the end of the chapter, we will learn about the policy gradient algorithm called the Distributed Distributional Deep Deterministic Policy Gradient (D4PG). We will learn what the D4PG...