Support Vector Machines for Classification: A Statistical Portrait

The support vector machine is a supervised learning technique for classification increasingly used in many applications of data mining, engineering, and bioinformatics. This chapter aims to provide an introduction to the method, covering from the basic concept of the optimal separating hyperplane to its nonlinear generalization through kernels. A general framework of kernel methods that encompass the support vector machine as a special case is outlined. In addition, statistical properties that illuminate both advantage and limitation of the method due to its specific mechanism for classification are briefly discussed. For illustration of the method and related practical issues, an application to real data with high-dimensional features is presented.