Mathematical Pictures at a Data Science Exhibition

 The book discusses a list of several representative algorithms used in data science today. However, the coverage is in-depth. The author discusses several mainstream machine learning techniques, including classification, support vector machines, regression, neural networks, and clustering. The compressive sensing chapter is an extension of the excellent previous book by the author and covers the increasingly important area of dealing with large data sets. The section on optimization is another gem that sometimes receives less attention in the data analytics community. 

One of the most important chapters is the one on optimal recovery techniques. These techniques focus on algorithms that compute the worst-case scenario error instead of error generalization techniques and deal with data points in a deterministic fashion rather than random. Simon Foucart discusses the mathematical theory of optimal recovery techniques, approximability models, ways to handle dimensionality problems, and quasi-Monte Carlo methods. The author provides an excellent review of relevant literature for the audience that needs broader coverage of this topic. The reference to the integration of trigonometric polynomial is a great find.