The Master Algorithm (2015) by Pedro Domingos looks at machine learning and describes the possible impact of machine learning on society and provides a survey for layman of major methods used in machine learning. Domingos is a Professor at the University of Washing in machine learning who also has an online course for learning more about Machine Learning on Coursera.
The book is perhaps a bit too keen in boosting machine learning, but it may be that the impact of machine learning is going to be as big as Domingos predicts. However, even if it’s somewhat less machine learning has already had a substantial impact in Search, recommendations and autonomously operation vehicles.
The book divides the machine learning world into five camps, evolutionist, connectionists, symbolists, Bayesians and analogizers. The symbolists use inverse deduction, the evolutionists genetic algorithms, the connectionists back propagation, the Bayesians, Bayesian classifiers, the analogizers use support vector machines.
Domingos then describes how what he researching, Markov logic networks are going to provide a unifying system for all of machine learning. Domingos also provides a short an interesting critique of the singularity and machine learning.
The book is fun to read if you’re a programmer and are looking for some inspiration to learn some machine learning and are looking to get some awareness of machine learning. It’s inspiring enough for me to make me want to go and learn more and actually program and play with some of the algorithms described.