This is not my favorite machine learning book, but Tom Mitchell did us all a favor by writing it. It covers the breadth of topics that make up the machine learning discipline fairly completely. Since this book is about completely, there is also a shallowness, but that shallowness does not trim out complete descriptions of the algorithms covered. Oh no, all the gory math is there, what isn't there are simple examples.

My first time through the book, what gave me the biggest headache was trying to understand back propagation from the algorithm pseudo code and the proof of correctness. I really really wanted one simple example at that point to make sure I understood the correct use of all the greek terms.

So good book, but I really wanted "Machine Learning Examples" to go along with it back when I first picked it up. But once you understand, the book is a great reference.

The book machine learning by Mitchell provides a systematic overview of important concepts in the field. This is rather rare finding because most books present first of all algorithms but fall short communicating systematic insights that would help the reader to creatively develop methods by themselves.


It is needless to say that any book with the title 'machine learning' is inherent incomplete due to the incompletenss of the field itself. For this reason this book is not state of the art of current algrithms. Instead, again, concepts are at the center of focus.

Overall, well writen and a very good selection of examples and explanations. I recommend this to anyone for a general overview.

This book is a redaction of many different white papers on the topic of machine learning. The material is very credible and accepted in the field, with very little (if any) temporal information (short term at least). With that said, it is also very dry and academic, and requires a solid background in mathematics to understand. Even if you are in the field, you're likely to read certain pages several times to embrace a concept... but once you embrace it, you will have some of the best foundational knowledge there is on the subject. If you're in the machine-learning field, you'll benefit from revisiting some of these subject, and probably learn a new thing or two.

I read this book about 7 years ago while in the PhD program at Stanford University. I consider this book not only the best Machine Learning book, but one of the best books in all of Computer Science. It covers every branch of ML I know of and it covers it really well. I found Mitchell's chapter on Neural Networks more insightful than an entire book on NN's that I read. I also found his chapter on Reinforcement Learning more useful and better explained than an entire book on Reinforcement Learning that I also read. The other chapters cover other ML topics at the same level of quality and rigor.

The author did an amazing job in covering the breadth and depth of ML in less than 500 pages. I hope he will write a new edition to cover the advances that happened in the last decade.

I have used this book during my masters and found it to be an extremely helpful and a gentle introduction to the thick and things of machine learning applications. The various chapters are nicely paced with helpful problems at the end. Another great thing about the book is treatment of detailed examples with each concept and that the author carefully ties various concepts as they arise, with not just new, but also examples from previous chapters, which helps the user to understand different concepts applied to same problems thereby making clear difference between different methods. Also the author has a dedicated website with updated errata and notes, which is also very helpful! Having said that, I think the book is an introduction to various machine learning methods and one can easily follow on the references listed for detailed treatment of relevant topics.