I. Introduction
Support VECTOR MACHINE (SVM) was first introduced to solve the pattern classification and regression estimation problem by Vapnik and his colleagues [1]–[3]. It can be seen as a new training algorithm for the traditional polynomial, radial basis function, and multilayer perceptron classifier by defining relevant kernel functions. In this paper, we have named it the standard SVM training algorithm. The main idea of SVM is to derive a hyperplane by maximizing the margin between two classes. The interesting property of SVM is that it is an approximate implementation to the structure risk minimization (SRM) principal in statistical learning theory, rather than the empirical risk minimization method, in which the classification function is derived by minimizing the mean square error (MSE) over the data set. In recent years, it has been found in a significant amount of literature that SVM leads to remarkable improvements in handwritten digit recognition [1], image classification, and face detection [4]–[6], object detection [7], text categorization, and nonlinear time-series prediction [8].