This book teaches computational intelligence (CI) in a thorough, methodological manner that is theoretically profound and educationally oriented. The authors present state-of-the-art, real-world CI problems within a technical scope that reflects the latest advancements of the field. The content is informative, and the explanations of the basics of CI are easy to comprehend.

The book is divided into four parts: “Neural Networks,” “Evolutionary Computing,” “Fuzzy Systems,” and “Bayes Networks.” Each part includes five to eight chapters, depending to the range of the topic. The section on neural networks is the major part of the book, with eight chapters that focus on basic neuron models such as threshold logic units, perceptrons, feedforward networks, recurrent networks, Hopfield networks, Kohonen networks, and radial basis functions. This part concludes with a chapter on some of the mathematical aspects essential to neural network learning.

The second part considers evolutionary systems. The authors discuss the biological inspiration behind the development of evolutionary algorithms, and their application in solving optimizations such as the traveling salesman problem. This section also explores other interesting methods and applications, such as genetic programming, which tries to create and evolve computer programs by associating symbolic expressions, swarm intelligence, and self-organization, based on the evolution and cooperation of agents that constitute a population. The agents of the population cooperate to achieve one target by following simple rules. In this area, intelligence is neither centered nor agent oriented, although the cooperation of all the agents in the population exhibits an intelligent behavior.

Fuzzy systems bypass the traditional methodology for building inferences, which uses predicates and propositions that assign either a true or a false value to an entity. Fuzzy systems are instead based on the degree of truthfulness of an entity, which ranges from 0 to 1. Fuzzy logic is introduced in the first chapter of this part. Other chapters explore fuzzy sets, fuzzy relations, fuzzy control, and fuzzy clustering. Note that the fuzzy control and fuzzy clustering approaches combine neural networks learning, evolutionary algorithms, and clustering within fuzzy systems.

The last part of the book focuses on Bayes networks, which are based on probability theory. Such networks are formally known as directed acyclic graphs, composed from vertices and edges. The vertices represent random variables and the edges correspond to the dependency of one variable to another. The first two chapters of this part introduce Bayes networks, with an emphasis on probability and graph theory. The third chapter explains dependence graphs and independence graphs, with a real-world application involving knowledge representation. The application supports the extraction of “expert knowledge representation” through a car manufacturing process. The latter is achieved using a Bayes network, specifically the Markov model, that is applied to the database of a car manufacturing company (Volkswagen).

The remaining chapters discuss the K2 algorithm, which enables a Bayes network to learn and induce structure from data, and propagation algorithms, which propagate new information through a system and build new inferences when new data is added to the system.

I found this book interesting because it provides a mathematical background for each chapter that is very clear and very well shaped. This provides added value compared to other books that treat the same topics but minimize the mathematical part, which is essential for understanding the subject and applying it to similar research problems.

I believe this book is well designed for the independent student who wishes to learn the fundamentals of CI without the need for an instructor. The organization and thorough step-by-step methodology makes it an excellent startup guide for someone who wants to learn CI but has little or no background in the field. I know of no other books that can provide a similar scope and methodology.

This book is targeted at beginners, students, or professionals who wish to understand CI. Other books [1] might be a better choice for researchers and advanced students who already have a background in CI, but want to deepen and broaden their knowledge and expertise in the field.