Book Image

TensorFlow Reinforcement Learning Quick Start Guide

By : Kaushik Balakrishnan
Book Image

TensorFlow Reinforcement Learning Quick Start Guide

By: Kaushik Balakrishnan

Overview of this book

Advances in reinforcement learning algorithms have made it possible to use them for optimal control in several different industrial applications. With this book, you will apply Reinforcement Learning to a range of problems, from computer games to autonomous driving. The book starts by introducing you to essential Reinforcement Learning concepts such as agents, environments, rewards, and advantage functions. You will also master the distinctions between on-policy and off-policy algorithms, as well as model-free and model-based algorithms. You will also learn about several Reinforcement Learning algorithms, such as SARSA, Deep Q-Networks (DQN), Deep Deterministic Policy Gradients (DDPG), Asynchronous Advantage Actor-Critic (A3C), Trust Region Policy Optimization (TRPO), and Proximal Policy Optimization (PPO). The book will also show you how to code these algorithms in TensorFlow and Python and apply them to solve computer games from OpenAI Gym. Finally, you will also learn how to train a car to drive autonomously in the Torcs racing car simulator. By the end of the book, you will be able to design, build, train, and evaluate feed-forward neural networks and convolutional neural networks. You will also have mastered coding state-of-the-art algorithms and also training agents for various control problems.
Table of Contents (11 chapters)

Learning the Markov decision process

The Markov property is widely used in RL, and it states that the environment's response at time t+1 depends only on the state and action at time t. In other words, the immediate future only depends on the present and not on the past. This is a useful property that simplifies the math considerably, and is widely used in many fields such as RL and robotics.

Consider a system that transitions from state s0 to s1 by taking an action a0 and receiving a reward r1, and thereafter from s1 to s2 taking action a1, and so on until time t. If the probability of being in a state s' at time t+1 can be represented mathematically as in the following function, then the system is said to follow the Markov property:

Note that the probability of being in state st+1 depends only on st and at and not on the past. An environment that satisfies the following state transition property and reward function as follows is said to be a Markov Decision Process (MDP):

Let's now define the very foundation of RL: the Bellman equation. This equation will help in providing an iterative solution to obtaining value functions.