Title: On proximal bilateral-weighted fuzzy support vector machine classifiers
Authors: S. Balasundaram; M. Tanveer
Addresses: School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India ' School of Computer and Systems Sciences, Jawaharlal Nehru University, New Delhi 110067, India
Abstract: A new approach for classification problems, called proximal bilateral-weighted fuzzy support vector machine, is proposed wherein each input example is treated as belonging to both positive and negative classes with different fuzzy memberships. The assumption of treating every input example belonging to both the classes is very well justified in real world applications. For example, for the study of credit risk assessment a customer can not always be assumed to be absolutely good or bad as he may default or pay his debit at times and therefore he may be treated as belonging to both the classes. Our formulation leads to solving a system of linear equations of size equals to the number of input examples. Computational results of the proposed method on publicly available datasets including two credit risk analysis datasets to that of the standard, proximal and bilateral-weighted fuzzy support vector machine methods clearly demonstrates its efficiency and usefulness.
Keywords: bilateral-weighted fuzzy SVMs; support vector machines; B-FSVMs; credit risk; risk assessment; fuzzy membership; proximal SVMs; PSVMs; SVM classifiers; classification.
International Journal of Advanced Intelligence Paradigms, 2012 Vol.4 No.3/4, pp.199 - 210
Received: 26 Sep 2011
Accepted: 23 Jul 2012
Published online: 23 Aug 2014 *