I teach several classes on machine learning (ML), at both the undergraduate and graduate levels. I am always looking for books that may provide new examples or ideas on how to teach ML to my students. The diversity in their knowledge of mathematics and statistics is high, which makes teaching a challenge. Thus, I was quite interested in how the author spans this gap. He claims to provide an informal and rigorous introduction to, and description of, the area of deep learning.

The book covers all the main topics that are associated with deep learning, from artificial neural networks to reinforcement learning, which is quite a hot topic. Readers with a good mathematical background can read the chapters, at least partially, independent from each other. I like that all the formulas are explained. The author also provides pseudocode, which delivers the main mathematical descriptions in a more computational way, as well as Python implementations. Although there is a lot of source code available for ML and for professional use, readers may prefer using TensorFlow or Scikit-learn for implementation help to follow the book, although most of the code is just modified versions from other web resources. The appendices contain a short introduction to matrices, ML estimation, and Bayes’ theorem. This is quite helpful for readers who have never heard about these topics and need more in-depth explanations.

This is a good book for students and anyone interested in ML; however, to read and grasp the topics quickly, basic university-level mathematics is required. As a further evaluation, I will ask my students to read selected chapters that I do not cover in my lectures and see whether they successfully provide an informal but rigorous introduction.

More reviews about this item: Amazon