Beginning Reinforcement Learning

Developers teaching software agents to learn from their environment, from walking robots, talking avatars and playing games, to rewriting programs.About This Book* A no-holds-barred introduction to reinforcement learning from first principles to the latest and greatest algorithms*How to implement fresh RL algorithms and make them part of your project*Learn the boundaries and applications of an area so new that algorithms and approaches are invented every monthWho This Book Is ForFluency in Python is assumed. Basic deep learning approaches should be familiar, and some practical experience in deep learning is helpful. This book is meant to be an introduction to reinforcement learning (RL), and requires no background in RL.What You Will Learn* How to understand the deep learning context of reinforcement learning*How to implement simple RL techniques like the Bellman equation*Applying Policy Gradient approaches to the real world*Defeat computer games without ever touching a keyboard*Learn the required deep learning and machine learning methods to understand reinforcement learningIn DetailReinforcement Learning (RL) is much more than the newest buzzword in deep learning. Like most areas in machine learning, the first popular texts have been around since the late 90s, but it is only since Google started to use reinforcement learning algorithms to play and defeat well-known computer games, that the field shot to prominence.This is the first book presenting reinforcement learning (RL) from first principles. It presents RL algorithms and methods developed since the late 90s, in an accessible and practical fashion. RL stands for the art of coding intelligent learning agents able to adapt to a formidable array of tasks.Max Lapan leads through some well-known areas like the Bellman equation and dynamic programming, and also introduces Deep-Q Network problems and Policy Gradient approaches in some depth. Max ends with a ride through some of the recent developments in RL, suggesting applications and new departures.