Book Image

Mastering Machine Learning Algorithms

Book Image

Mastering Machine Learning Algorithms

Overview of this book

Machine learning is a subset of AI that aims to make modern-day computer systems smarter and more intelligent. The real power of machine learning resides in its algorithms, which make even the most difficult things capable of being handled by machines. However, with the advancement in the technology and requirements of data, machines will have to be smarter than they are today to meet the overwhelming data needs; mastering these algorithms and using them optimally is the need of the hour. Mastering Machine Learning Algorithms is your complete guide to quickly getting to grips with popular machine learning algorithms. You will be introduced to the most widely used algorithms in supervised, unsupervised, and semi-supervised machine learning, and will learn how to use them in the best possible manner. Ranging from Bayesian models to the MCMC algorithm to Hidden Markov models, this book will teach you how to extract features from your dataset and perform dimensionality reduction by making use of Python-based libraries such as scikit-learn v0.19.1. You will also learn how to use Keras and TensorFlow 1.x to train effective neural networks. If you are looking for a single resource to study, implement, and solve end-to-end machine learning problems and use-cases, this is the book you need.
Table of Contents (22 chapters)
Title Page
Dedication
Packt Upsell
Contributors
Preface
13
Deep Belief Networks
Index

SARSA algorithm


SARSA (whose name is derived from the sequence state-action-reward-state-action) is a natural extension of TD(0) to the estimation of the Q function. Its standard formulation (which is sometimes called one-step SARSA, or SARSA(0), for the same reasons explained in the previous chapter) is based on a single next reward, rt+1, which is obtained by executing the action at in the state st. The temporal difference computation is based on the following update rule:

The equation is equivalent to TD(0), and if the policy is chosen to be GLIE, it has been proven (in Convergence Results for Single-Step On-Policy Reinforcement-Learning Algorithms, Singh S., Jaakkola T., Littman M. L., Szepesvári C., Machine Learning, 39/2000) that SARSA converges to an optimal policy, πopt(s), with the probability 1, when all couples (state, action) are experienced an infinite number of times. This means that if the policy is updated to be greedy with respect to the current value function induced by...