Fuzzy Support Vector Machine Using Function Linear Membership and Exponential with Mahanalobis Distance

Support vector machine (SVM) is one of effective biner classification technic with structural risk minimization (SRM) principle. SVM method known as successful in technic. But the real-life data problem lies occurrence noise and outlier. Noise will create confusion for when being processed. On this research, developed by adding its fuzzy membership function to lessen outlier effect trying figure out hyperplane solution. Distance calculation also considered while determining value because it a basic thing proximity between elements, which general built depending on distance point into real class mass center. Fuzzy support (FSVM) uses Mahalanobis distances goal finding best separating defined classes. The used be going over trial several dividing partition percentage transforming training set testing set. Although theoretically FSVM able overcome outliers, results show that accuracy FSVM, namely 0.017170689 0.018668421, lower than classical method, 0.018838348. existence extremely influential deciding hyperplane. Based that, correct critical problem.