This first edition is a text for research students and practitioners whose interest goes beyond black-box solutions. It starts with an introduction to regression, classification, probability, and statistics, followed by detailed explanations of the classical mean square error estimation and least mean square algorithms. These classic topics are followed by more advanced subjects in machine learning such as convexity, sparse modeling, Bayesian learning, and Monte Carlo sampling methods, these being more tedious subjects for nonexperts to grasp. It eventually ends with probabilistic graphical models, neural networks, deep learning, and dimensionality reduction techniques.

The book is not suitable for undergraduate students facing machine learning for the first time, as it contains a significant amount of mathematical, statistical, and formal terms. The author provides a detailed and comprehensive picture of the techniques, approaches, and Bayes-based methodologies that can be useful for practicing scientists and engineers. A strength of the book is that each chapter is a self-contained piece of text that can be employed by an instructor or practitioner for his or her own teaching or research. A complete set of references follows each chapter, putting the presented material into context. Although the advanced chapters are followed by a case study and MATLAB exercises, a greater emphasis on the application of theoretical material would have been more beneficial to readers. The title includes “optimization perspective,” but this is a counterintuitive and misleading proposition. The book should clearly highlight those algorithms and techniques within each approach that cover optimization. I was expecting a conclusive chapter with an indication of open challenges and future directions in the fast-growing field of formal classification and prediction techniques.

Overall, this text is well organized and full of details suitable for advanced graduate and postgraduate courses, as well as scholars; however, it is not suitable for novices facing machine learning for the first time.

More reviews about this item: Amazon, Goodreads