Evolutionary Training of SVM for Multiple Category Classification Problems with Self-adaptive Parameters

  • Ángel Kuri-Morales
  • Iván Mejía-Guevara
Conference paper

DOI: 10.1007/11874850_37

Volume 4140 of the book series Lecture Notes in Computer Science (LNCS)
Cite this paper as:
Kuri-Morales Á., Mejía-Guevara I. (2006) Evolutionary Training of SVM for Multiple Category Classification Problems with Self-adaptive Parameters. In: Sichman J.S., Coelho H., Rezende S.O. (eds) Advances in Artificial Intelligence - IBERAMIA-SBIA 2006. Lecture Notes in Computer Science, vol 4140. Springer, Berlin, Heidelberg

Abstract

We describe a methodology to train Support Vector Machines (SVM) where the regularization parameter (C) is determined automatically via an efficient Genetic Algorithm in order to solve multiple category classification problems. We call the kind of SVMs where C is determined automatically from the application of a GA a “Genetic SVM” or GSVM. In order to test the performance of our GSVM, we solved a representative set of problems by applying one-versus-one majority voting and one-versus-all winner-takes-all strategies. In all of these the algorithm displayed very good performance. The relevance of the problem, the algorithm, the experiments and the results obtained are discussed.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Ángel Kuri-Morales
    • 1
  • Iván Mejía-Guevara
    • 2
  1. 1.Departamento de ComputaciónInstituto Tecnológico Autónomo de MèxicoMéxico
  2. 2.Posgrado en Ciencia e Ingeniería de la ComputaciónUniversidad Nacional Autónoma de México, IIMASMéxico