Deep training with reinforcement.alphago and other technologies

This book is a detailed guide to the latest deep learning tools with reinforcement and their restrictions. We will realize and check in practice the methods of cross-endropy and iteration on values ​​(Q-learning), as well as gradients in strategies.
The experiments use a variety of environmental training (RL), starting with classic Cartpole and Gridworld to ATARI emulators and continuous control media (based on Pybullet and Roboschool) Many examples are based on non -standard environments in which we will develop the environment from scratch.
In this book
- You will find out what place in the context of deep learning is occupied by RL methods, implement complex models of deep learning.
- Study the basis of RL: Markov decision-making processes.
- Consider examples of the implementation of RL methods: method of cross-entropy, DQN, A3C, TRPO, PPO, DDPG, D4PG and others.
- You will learn how to work with discrete and continuous spaces of actions in various