A fuzzy classification method based on support vector machine

Support vector machine (SVM) is a novel type learning machine, based on statistical learning theory. Due to the good generalization capability, SVMs have been widely used in classification, regression and pattern recognition. In this paper, for the data with numerical condition attributes and decision attributes, a new fuzzy classification method (FCM) based on SVM is proposed. This method first fuzzifies decision attributes to some classes (linguistic terms), then trains decision function (classifier). For a new sample, the decision function doesn´t forecast the value of its decision attribute, but gives the corresponding class and its membership degree as fuzzy decision. This fuzzy decision result is more objective and easier to understand than crisp decision in common sense. The design principle is given and the classification algorithm is implemented in this paper. The experimental results show that the new method proposed in this paper is effective. The method optimizes the classified result of common SVMs, and therefore enhances the intelligent level of SVMs.