Pattern Recognition and machine learning (preface translation)

Objective

Since machine learning is generated from computer science, image recognition originates from engineering. However, these activities can be seen as two aspects of the same field, and they have undergone a fundamental development in the past 10 years. In particular, when the image model has emerged as a framework for describing and applying probabilistic models, the Bayesian theorem (Bayesian methods) has evolved from an expert-level knowledge category to a mainstream one. Through a series of approximate algorithms, such as variable decibel Dean and desired propagation (variational Bayes and expectation propagation), the practical application scope of Bayesian theorem has been greatly improved. At the same time, the new model based on core program has important influence on algorithm and application.

This new book provides a comprehensive introduction to image recognition and machine learning, and it also reflects the current state of development. It is designed primarily for excellent graduate students or first-year doctoral students, as well as for researchers and practitioners, and is based on the assumption that there is no learning experience in image recognition and machine learning concepts. Of course, multivariate calculus and basic linear algebra are needed, and a certain degree of mastery of probability theory will be helpful, although there is no mandatory requirement because the book itself contains an introduction to the underlying probability theory.

Since this book covers a wide range, it is not possible to provide a complete derivation process, and it is not intended to specifically introduce the precise historical attribution of the concept. Instead, our goal is to give references, which provide the greatest possible detail, and in some cases wish to provide an entry point for this very broad subject matter. For this reason, most of the references are current textbooks and commentary articles rather than raw resources.

This book cites a lot of extra information, including course slides and icons that are completely used in textbooks. and encourage readers to go to the book website to get the latest information: HTTP://RESEARCH.MICROSOFT.COM/∼CMBISHOP/PRML

Exercises

The problem that appears at the end of each section is an important part of the book. Each exercise is carefully selected to enhance the concept of interpretation in the text, or in a meaningful way to develop and promote them, and each problem is divided into a star to Samsung, a star represents a simple exercise, only a few minutes to complete; Samsung represents a significantly more difficult problem.

It is difficult to know which exercises range to give the answer to the broad benefit of everyone. Self-directed learners will find that the ready-made answers are very helpful, and many course teachers request only the answers given by the publishers to be better, because these exercises can be used in the classroom. In an effort to meet this conflicting requirement, those exercises to help detail key points in the text or to supplement important details will have ready-made answers, which are published on the website of the book in a PDF file. For the rest of the exercise answer the teacher can be contacted by the publisher (contact information posted on the book website). Readers are strongly encouraged to solve these problems without being helped, only to see the answers if necessary.

While the book focuses on concepts and principles, the idea is that students should have the opportunity to experiment with the key algorithms in the appropriate set of numbers in class. A sister article (Bishop and nabney,2008) will address the practice aspects of image recognition and machine learning, and will use MATLAB software to practice most of the algorithms discussed in the book.

Thank

First I will be sincere thanks to Markus Svens´en for his great help in charting and typesetting the book. His assistance is priceless.

I am very thankful for Microsoft Research, because it provides me with a highly facilitated study environment and gives me the freedom to write this book (the book's Views and insights are only personal and irrelevant to Microsoft and its affiliates).

Springer is very supportive of the final stage of the book's preparation, and I would appreciate the support and professionalism of the appointed editor John Kimmel. At the same time, Joseph Piliero's cover design and the article format, Maryann Brickner Many production links to help express thanks. This cover design is inspired by discussions with Antonio Criminisi.

I would also like to thank the Oxford University Publishing (Oxford University Press) for a reference to a previously published book Neural Networks for Pattern recognition (bishop,1995a). The copy of the Mark 1 Perceptron and the Frank Rosenblatt picture has been Arvin Calspan Advanced Technology Center's permission. I would also like to thank Asela Gunawardana in Figure 13.1 for me to draw the spectra, while thanking Bernhard Sch¨olkopf allow me to use his core code PCA to draw 12.17.

Many people have provided assistance in proofreading draft materials and providing advice and advice, including Shivani Agarwal, C´edric Archambeau, Arik azran,andrew Blake, Hakan Cevikalp, Michael Fourm An, Brendan Frey, Zoubin Ghahramani, Thore Graepel, Katherine Heller, Ralf Herbrich, Geoffrey Hinton, Adam Johansen, MATTH EW Johnson, Michael Jordan, Eva Kalyvianaki, Anitha Kannan, Julia Lasserre, David Liu, Tom Minka, Ian Nabney, Tonatiuh Pen A, Yuan Qi, Sam Roweis,balaji Sanjiya, Toby Sharp, Ana Costa e Silva, David spiegelhalter, Jay Stokes, Tara symeonides, Ma Rtin Szummer, Marshall tappen, Ilkay Ulusoy, Chris Williams, Johnwinn, and Andrew Zisserman.

Finally, thanks to my wife, Jenna, she strongly supported me through the years of writing this book.

Chris Bishop

Cambridge

February 2006

PS: My younger brother first translation, and non-professional English, all kinds of mistakes and wrong to hope that you are the master, this is the book's preface. Thank you for taking the time to watch and support. Who has a more genuine PDF can send me a best, grateful.

Pattern Recognition and machine learning (preface translation)