Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond (Adaptive Computation and Machine Learning)
 von Bernhard Scholkopf
ISBN: 0262194759  more than an introduction If you would like to know what pattern recognition is, and how to solve this kind of problems with standard stateoftheart tools from statistical learning theory or with a bayesian approach, this is the book you can start with, even if you are not familiar with this technics.This splendid book is easy to read, starts with a simple introduction and continues with basic concepts and tools, like kernels,risk minimization,regularization, elements of statistical learning theory and optimization. Thereafter they discribe the support vector machine in a detail manner one of the stateoftheart tools for regression and classification. And they finish with kernel methods, like the kernel pca, kernel fisher discriminant, bayesian kernel methods and preimages. So it collects results, theorems, algorithms and discussions from different sources into one very accessible exposition. It is also a good reference book. Siehe auch: > Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond (Adaptive Computation and Machine Learning) 
