Many Classical ML Books - Enjoy and Do Stuffs

There are so many fancy books in machine learning field…Personally, I believe the following books are the most classical~~

Pattern Recognition And Machine Learning NOTE: This book has very high prestige in the ML field today! But not difficult to read and very helpful.

The Elements of Statistical Learning NOTE: With this book, we no longer need the other ML textbooks, including abstruse statistical inference, matrix and numerical algorithms, convex optimization mathematics. ESL and PRML focus different thing. The former perspective is frequentist and more intuitive, the latter perspective is Bayesian.

Bayesian Reasoning and Machine Learning NOTE: It is a new classical ML book. You can understand the basic concepts and algorithms of probabilistic graphical models just from its first chapter. It is logical, simple, clear and good layout. Also its MATLAB code package is very nice just like this book.

Convex Optimization NOTE: This book has very high prestige in the ML field today! But not difficult to read and very helpful.

Graphical Models, Exponential Families, and Variational Inference NOTE: Jordan’s classical work. It shows his deep thinking, and it is the best review in this area.

Multiple View Geometry in Computer Vision NOTE: Hartley’s this amazing book feed countless doctors and professors. There are part codes of this book in ox’s vgg website. If you are interesting in it, also you can read “In Defense of the Eight-Point Algorithm” published in 1997. This book is very rigorous and its algorithms are not difficult to understand. And the best part is the pseudo-code algorithms that is very, very helpful and understanding~

Numerical Optimization NOTE: Both two authors are SIAM Follow, from CS professional major and enjoy a high status in the field of optimization. This theoretical book is mainly about continuous optimization from the most simple method of Newton, gradient, constrained optimization’s dual, KKT and simplex method, smoothly, to the penalty function and interior point method in non-linear areas. Not only about mathematics, but the development history in the optimization field. In addition to the content, this book’s English language is also worthy of praise. At least with it I am relaxed and happy.

Paradigms of Artificial Intelligence Programming NOTE: Use the most natural notation available to solve a problem. If you directly solve problem by writing the intuitional Lisp code, you will not discover the advantage of Lisp. The strengths of Lisp is to build the problem model firstly before write an interpreter for the model. Many Lisp books are talking about its basic techniques, but not the practical applying method. It is difficult to play the power of Lisp if you only know its techniques such as grammar, libraries, macros…(This book has more than 900 pages with lots if Lisp code and modest illustrations)

Learn From Data NOTE: The textbook of Caltech Professor Yaser Abu-Mostafa. It is very intelligible. Also you can see relative videos in http://www.youtube.com/watch?v=mbyG85GZ0PI&list=PLD63A284B7615313A $28 Learning Theory in plain English reread in 8 hours

Ggplot2. elegant graphics for data analysis NOTE: R graphics is really profound, which absolutely is the mainstream language for drawing. Ggplot is one dazzling package in R. This faithful book shows the author Hadley Wickham has a deep understanding in data analysis and data visualization.

Introduction to Information Retrieval NOTE: The classic introductory book of Information Retrieval with clear and concise logic. You will learn a lot about search engine and simple NLP and so on in a short time.

Programming Collective Intelligence NOTE: This short book amazingly shows many basic algorithm theories and python project code. And you would better study it with some theoretical books to help you understand these and more algorithms.