LINEAR INDEPENDENCE AND SUPPORT VECTOR MACHINES
Dr Kevin Gates, Department of Mathematics, University of Queensland, Australia

Abstract: In this talk we will be discussing the concept of linear independence of vectors and show how the linear dependence of the support vectors can be used to reduced the number of vectors. After a basic discussion of the concept of linear independence we will show how post processing can considerably reduce the number of vectors required. The talk will conclude with the application of the concept to the set of training vectors and we will show how the set can be reduced in size with associated increases in efficiencies without significant losses in accuracy.

This seminar was held at the Department of Computer Science, Royal Holloway, University of London on 3 October 2002.