If you find any typos or have suggestions for improving the book, do not hestitate to contact us via email at: hao.dong[at].pku.edu.cn
The first part of this book has six chapters to introduce the foundations of deep learning (DL), reinforcement learning (RL), widely used DRL algorithms and their implementations. Specifically, the first two chapters introduce the basic knowledge ofDL and RL and the combination of the two, i.e. DRL, which are important for the readers to understand the rest of the book. You can skip these two chapters if you already have the related knowledge, but we highly recommend that you read through the second chapter to get familiar with the terminology and the mathematical formulas for the convenience of reading the following chapters. The third chapter introduces the taxonomy of RL algorithms, which is intended to help readers to have an overview of modern DRL algorithms from different perspectives, such as model-based and model-free, policy-based and value-based, MC and TD methods, on-policy and off-policy, etc. We recommend that the readers go back to this chapter if there is any confusion about the categories and properties of specific algorithms when reading other chapters. For specific DRL algorithms, we introduce those which are most commonly applied, in detail, from the fourth to sixth chapters as well as providing the example codes to help the readers to understand the details of the algorithms and their implementations. The related codes are released in the following link: https://github.com/tensorlayer/tensorlayer/tree/master/examples/reinforcement_learning.