A New Method for Classifying Random Variables Based on Support Vector Machine

Abstract

In this paper, a new version of Support Vector Machine (SVM) is proposed which any of training samples are considered the random variables. Hence, in order to achieve robustness, the constraint in SVM must be replaced with probability of constraint. In this new model, by applying the nonparametric statistical methods, we obtain the optimal separating hyperplane by solving a quadratic optimization problem. Afterwards, we present the least squares model of our proposed method. The efficiency of our proposed method is shown by several examples for both cases (linear and nonlinear) with probabilistic constraints.