Using support vector machines in fuzzy classification

Fuzzy classification is one of methods used for pattern classification, which is germane to many engineering applications. An output from fuzzy classification is an assignment of patterns to fuzzy classes. There are several methods for fuzzy classification; in this paper we propose a new method based on the soft margin support vector machines classifier (C-SVM). These classifiers are based on statistical learning theory and they are widely used in pattern classification. In our approach, the decision boundary and slack variables obtained from C-SVM are used for the definition of a new optimization problem. The goal is to find an optimal parametrized transformation function T , which transforms the distance of a pattern from de-cision boundary to its membership degree. Quadratic programming is used to find suitable values for the parameters of T . The development of the full scope of this new fuzzy classification method is still in progress.