Machine Learning recommendation Book list

Machine learning is often in dealing with maths, so reading a book is certainly essential. Here are some of the books that I have read and find helpful, and I hope it will be helpful to everyone. (Please ignore the bad typography, this typesetting function is too difficult to use.) ）

Topology:

Munkres J R. "Topology"
Topology only read this one, can only be said to be worthy of the master's work.
Analytical studies:

Rudin three-piece set
This should be well known. In addition, I read a series of calculus textbooks in crown early, which is very good for getting started. In addition, Crown's mathematical physics method is also a rare masterpiece, many of the perspective of dealing with problems are only seen in this book, but unfortunately only read the first volume of the previous chapters.

linear algebra:
Michael Artin. "Algebra"
Although this book is not a monograph of linear algebra, it is strongly strongly recommended as a textbook of linear algebra, which reads Chafanbusi, Yibingbuqi.

Peter D.lax "Linear Algebra and its Applications"
Another excellent textbook of linear algebra, the content is more extensive.

Probability statistics:
Dimitri P.bertsekas, John n.tsitsiklis "Introduction to Probability"
A relatively easy-to-understand probability theory textbook
Christian P. Robert, George Casella "Monte Carlo statistical Methods"
The application of Monte Carlo method in machine learning should be needless to say in the monograph of Monte Carlo method. Read a few similar textbooks, the feeling is similar, also said not good which is more suitable.

Optimization Method:
Stephen Boyd "Convex optimization"
Convex optimization in the field of Super Classic, the convex optimization of all aspects of the detailed narration, there is a Chinese version.

differential Geometry:
Marian Fecko "differential Geometry and Lie Groups for physicists"
When reading, the tutor recommended the textbook, mainly about modern differential geometry. This book is super thick, I only read the previous seven or eight chapters, benefited a lot, the understanding of linear algebra and calculus has deepened.

Numerical calculation:
William H. Press and other "numerical Recipes"
Several numerical calculations Daniel co-authored encyclopedia-style textbooks, not only in the field of numerical computing in detail, but also comes with high-quality source code, a lot of programs can be directly used. Of course, the book is very thick (1000+), but after reading through it should basically be able to deal with most of the problems encountered in the work of the numerical calculation.

Gene H. Golub "Matrix computation"
This should be done without too much introduction to the Bible in the field of matrix computing. To learn more about how to solve linear algebraic equations, to find the eigenvalues, the various matrix decomposition algorithm students must not be missed. In addition, there is a slightly thinner textbook, is written by the author of Lapack, but also very good.

Jorge Nocedal, Stephen J. Wright "Numerical optimization"
This book is an introduction to the numerical method to solve the optimization problem of monographs, can be combined with the convex optimization of Boyd, a partial theory of a partial practice

Machine Learning Series:

Kevin P. Murphy "machine learning:a Probabilistic Perspective"
Abbreviation Mlapp, is also I study machine study of the first book, is a chatty of books. can help beginners to quickly build a complete framework of machine learning content, to avoid falling into such specific algorithms as logistic regression, support vector machine, trees trees. However, due to space constraints, many chapters of the discussion is relatively simple, such as probability map model, Gaussian process, Dirichlet process, deep learning, etc., it is recommended to study together with related monographs.

Robert E. Schapire,yoav Freund "Boosting foundations and Algorithms"
As can be seen from the title, this is a book devoted to the boosting algorithm. Explains in many ways why the boosting algorithm is so awesome.

Mehryar Mohri, Afshin Rostamizadeh, and Ameet Talwalkar "Foundations of machine learning"
PAC Learning theory Monographs, but less used in the work, but also only about the understanding of the

Ian Goodfellow,yoshua bengio,aaron Courville "deep learning"
It doesn't have to be a lot of words ...

Andreas Griewank,andrea Walther "Evaluating derivatives:principles and techniques of algorithmic differentiation"
This book is about automatic differentiation, and it seems that few people recommend it, but the quality of the content is pretty good. After reading it should be able to really know what is the BP algorithm, and why the deep learning framework to adopt the BP algorithm. It is recommended to implement the forward and posterior automatic differential algorithms in order to deepen understanding and memory.
In addition, there is a more famous book "The Art of differentiating computer Programs-an Introduction toalgorithmic differentiation", but I did not see much, So I don't make too many comments.


Hangyuan Li "Statistical learning method"
Many Daniel have recommended this book to prove its quality. Content to introduce the main machine learning algorithms, mathematical derivation is rich and informative, can be combined with Mlapp to see.

Daphne koller,nir Friedman "probabilistic graphical Models-principles and techniques"
The Encyclopedia of Probability graph theory, the content of some storytellers is very scattered, not recommended to read. But personally think is a rare good book, the probability of the theory of the context of the explanation clearly clear. After reading it then to learn hmm, CRF, Kalman filter These specific algorithms have a kind of know it and know the reason why the feeling. Mlapp the chapters in the probability graph are heavily consumed by this book. I have time to get my reading notes sorted out later.

Carl Edward Rasmussen, Christopher k. Williams "Gaussian Processes for machine learning"
A monograph on the Gauss process, which is quoted extensively in the Mlapp Gauss process section.

Richard S. Sutton,andrew G. Barto "Reinforcement Learning-an Introduction"
Intensive learning of the introductory books, did not read, because the work is not applied to the reinforcement of learning ...

Machine Learning recommendation Book list