Book Image

Deep Reinforcement Learning Hands-On - Second Edition

By : Maxim Lapan
Book Image

Deep Reinforcement Learning Hands-On - Second Edition

By: Maxim Lapan

Overview of this book

Deep Reinforcement Learning Hands-On, Second Edition is an updated and expanded version of the bestselling guide to the very latest reinforcement learning (RL) tools and techniques. It provides you with an introduction to the fundamentals of RL, along with the hands-on ability to code intelligent learning agents to perform a range of practical tasks. With six new chapters devoted to a variety of up-to-the-minute developments in RL, including discrete optimization (solving the Rubik's Cube), multi-agent methods, Microsoft's TextWorld environment, advanced exploration techniques, and more, you will come away from this book with a deep understanding of the latest innovations in this emerging field. In addition, you will gain actionable insights into such topic areas as deep Q-networks, policy gradient methods, continuous control problems, and highly scalable, non-gradient methods. You will also discover how to build a real hardware robot trained with RL for less than $100 and solve the Pong environment in just 30 minutes of training using step-by-step code optimization. In short, Deep Reinforcement Learning Hands-On, Second Edition, is your companion to navigating the exciting complexities of RL as it helps you attain experience and knowledge through real-world examples.
Table of Contents (28 chapters)
26
Other Books You May Enjoy
27
Index

Value, state, and optimality

You may remember our definition of the value of the state from Chapter 1, What Is Reinforcement Learning?. This is a very important notion and the time has come to explore it further.

This whole part of the book is built around the value and how to approximate it. We defined the value as an expected total reward (optionally discounted) that is obtainable from the state. In a formal way, the value of the state is , where rt is the local reward obtained at step t of the episode.

The total reward could be discounted with or not (the undiscounted case corresponds to ); it's up to us how to define it. The value is always calculated in terms of some policy that our agent follows. To illustrate this, let's consider a very simple environment with three states:

  1. The agent's initial state.
  2. The final state that the agent is in after executing action "right" from the initial state. The reward obtained from this is 1.
  3. ...