Book Image

Deep Reinforcement Learning with Python - Second Edition

By : Sudharsan Ravichandiran
Book Image

Deep Reinforcement Learning with Python - Second Edition

By: Sudharsan Ravichandiran

Overview of this book

With significant enhancements in the quality and quantity of algorithms in recent years, this second edition of Hands-On Reinforcement Learning with Python has been revamped into an example-rich guide to learning state-of-the-art reinforcement learning (RL) and deep RL algorithms with TensorFlow 2 and the OpenAI Gym toolkit. In addition to exploring RL basics and foundational concepts such as Bellman equation, Markov decision processes, and dynamic programming algorithms, this second edition dives deep into the full spectrum of value-based, policy-based, and actor-critic RL methods. It explores state-of-the-art algorithms such as DQN, TRPO, PPO and ACKTR, DDPG, TD3, and SAC in depth, demystifying the underlying math and demonstrating implementations through simple code examples. The book has several new chapters dedicated to new RL techniques, including distributional RL, imitation learning, inverse RL, and meta RL. You will learn to leverage stable baselines, an improvement of OpenAI’s baseline library, to effortlessly implement popular RL algorithms. The book concludes with an overview of promising approaches such as meta-learning and imagination augmented agents in research. By the end, you will become skilled in effectively employing RL and deep RL in your real-world projects.
Table of Contents (22 chapters)
Other Books You May Enjoy

Why distributional reinforcement learning?

Say we are in state s and we have two possible actions to perform in this state. Let the actions be up and down. How do we decide which action to perform in the state? We compute Q values for all actions in the state and select the action that has the maximum Q value. So, we compute Q(s, up) and Q(s, down) and select the action that has the maximum Q value.

We learned that the Q value is the expected return an agent would obtain when starting from state s and performing an action a following the policy :

But there is a small problem in computing the Q value in this manner because the Q value is just an expectation of the return, and the expectation does not include the intrinsic randomness. Let's understand exactly what this means with an example.

Let's suppose we want to drive from work to home and we have two routes A and B. Now, we have to decide which route is better, that is, which route helps us to reach...