Modified fuzzy support vector machine for credit approval classification – IOS Press

IOS Press c/o Accucoms US, Inc. For North America Sales and Customer Service West Point Commons 1816 West Point Pike Suite 125 Lansdale, PA 19446 USA For editorial issues, permissions, book requests, submissions and proceedings, contact the Amsterdam office: Inspirees International (China Office) Ciyunsi Beili 207(CapitaLand), Bld 1, 7-901 100025, Beijing China Free service line: 400 661 8717 For editorial issues, like the status of your submitted paper or proposals, write to editorial@iospress.nl Searching for just a few words should be enough to get started. If you need to make more complex queries, use the tips below to guide you. Faculty of Post Graduate Studies and Research, Computer Engineering and Technology, Marwadi Education Foundation's Group of Institutions, Rajkot, India. E-mail: arindam_chau@yahoo.co.in In the recent past, credit approval is a significant problem in credit risk management. Making decision to approve a credit has been a source of major concern for financial institutions. As such the problem is formulated as classification problem where making correct decision yields maximum returns . The classification task is taken care of by modified fuzzy support vector machine (MFSVM). It is variant of fuzzy support vector machine (FSVM) developed by Chaudhuri et al. The inherent vagueness and uncertainty in training samples are handled by new fuzzy membership function with hyperbolic tangent kernel. The success of classification lies in considering fuzzy membership function as function of center and radius of each class in feature space and representing it with kernel. In nonlinear training samples, input space is mapped into high dimensional feature space to compute separating surface using linear separating method. The different input points make unique contributions to decision surface. MFSVM produces significant results for Australian Credit Approval dataset. The model is tested with both linear and nonlinear kernels. MFSVM performance is also assessed in light of number of support vectors required to model the data. The effect of variability in prediction and generalization of MFSVM is studied with respect to parameters C and δ2. The area under curve helps to reduce imbalance issues in the dataset considered. The training samples are either linear or nonlinear separable. MFSVM effectively handles the issue of nonlinear classification problem. Experimental results on both artificial and real datasets support the fact that MFSVM achieves superior performance in reducing outliers' effects than FSVM. Corresponding author: Rui Ji, Department of Automation, Shanghai Jiao Tong University, Shanghai 200240, China. Tel.: +86 21 3420 4261, Fax: +86 … In this paper, we develop a novel support vector algorithm with fuzzy hyperplane for pattern classification. We first introduce the concepts of fuzzy … This work is supported by the National Natural Science Foundation of China (No. 11161045), the China Postdoctoral Science Foundation (No. 2015M572625)… Corresponding author. Carlo Vercellis, Department of Management, Economics and Industrial Engineering, Politecnico di Milano, Via Lambruschini 4b, 201… Corresponding author. Ai-bing Ji, Tel.: +86-25-84896481, Fax: +86-25-84498069, Emails: s.chen@nuaa.edu.cn (S. Chen), jabpjh@163.com (A.-b Ji). Suppor… Source.


Яндекс.Метрика Рейтинг@Mail.ru Free Web Counter
page counter
Last Modified: April 18, 2016 @ 2:05 pm