Fuzzy rules extraction from support vector machines for multi-class classification with feature selection

Abstract

Although Support Vector Machines (SVMs) have been successfully applied to many problems, they are considered "black box models". Some methods have been developed to reduce this limitation, among them the FREx-SVM, which extracts fuzzy rules from trained SVMs for multi-class problems. This work deals with an extension to the FREx-SVM method, including a wrapper feature subset selection algorithm for SVMs. The method was evaluated in four benchmark databases. Results show that the proposed extension improves the original FREX-SVM, providing better rule coverage and a lower number of rules, which is a considerable gain in terms of interpretability. © 2009 Springer Berlin Heidelberg.