SUPPORT VECTOR METHODS IN LEARNING AND FEATURE EXTRACTION
Dr Bernhard Schoelkopf, GMD FIRST, Berlin

Abstract: The last years have witnessed an increasing interest in Vapnik's Support Vector machines, which use Mercer kernels for efficiently performing computations in high-dimensional spaces. In pattern recognition, the Support Vector algorithm constructs nonlinear decision functions by training a classifier to perform a linear separation in some high-dimensional space which is nonlinearly related to input space. Recently, we have developed a technique for Nonlinear Principal Component Analysis based on the same types of kernels. This way, we can for instance efficiently extract polynomial features of order n by computing projections on to principal components in the space of all products of n pixels of images.

In the talk, I will explain the idea of Mercer kernels, and give an overview of the two above approaches. Application examples will be taken from the domain of pattern recognition.

This seminar was held at the Department of Computer Science, Royal Holloway, University of London on 17 November 1997.

back