This slim book is an excellent introduction to an exciting new field--the design and implementation of important new mathematical models as the optimising strategy for learning machines. The text deals clearly with both the concepts and the practical details of these models, and consistently steers away from the more speculative side of learning machines. At the same time, the authors never fail to communicate the wonder and excitement of working with the raw stuff of knowledge and impart a spark of intelligence to mechanisms.
An Introduction to Support Vector Machines is manifestly a text book, and as such leads the reader as a student through the concepts, history and implementation of kernel-based learning strategies, with plenty of pseudo-code examples, discussion and exercise questions (without answers), and the best modern bibliography of the subject available. The appendix attempts to summarise the background mathematics. It's thorough, accurate and useful as a reference: but it is not a tutorial, and may leave the novice reader with little training in set dynamics none the wiser. A well-rounded reader will sail through the intriguing first chapter, which discusses learning machines and techniques for teaching a machine (or computer program) to generalise. However chapter two, with its impenetrable conversation about "linear classification" replete with sigma notation and diagrams of "hyperplanes" may well discourage further reading. Excellent though this book is, the title is deceptive: it is indeed an "introduction"--but requires a fair background knowledge of automata and set theory.--Wilf Hey
'… the most accessible introduction to the area I have yet seen'. D. J. Hand, Publication of the International Statistical Institute
'The book is an admirable presentation of this powerful new approach to pattern classification.' Alex M. Andrew, Robotica
' … an excellent book, complete and readable without big requirements in mathematical functional analysis.' Zentralblatt für Mathematik und ihre Grenzgebiete Mathematics Abstracts