Choosing Multiple Parameters for Support Vector Machines

Unknown author(s)

2000 · 79 citations · 0 references

Concepts

Abstract

The problem of automatically tuning multiple parameters for pattern recognition Support Vector Machines (SVMs) is considered. This is done by minimizing some estimates of the generalization error of SVMs using a gradient descent algorithm over the set of parameters. Usual methods for choosing parameters, based on exhaustive search become intractable as soon as the number of parameters exceeds two. Some experimental results assess the feasibility of our approach for a large number of parameters (more than 100) and demonstrate an improvement of generalization performance. 1 Introduction In the problem of supervised learning, one takes a set of input-output pairs Z = f(x 1 ; y 1 ); : : : ; (x ` ; y ` )g and attempts to construct a classier function f that maps input vectors x 2 X onto labels y 2 Y. We are interested here in pattern recognition or classication, that is the case where the set of labels is simply Y = f1; 1g. The goal is to nd an f 2 F which minimizes the error (f(x) 6=...