Fuzzy Support Vector Regression Based on Data Domain Description

In order to overcome the overfitting problem caused by noises and outliers in support vector machines (SVM), a fuzzy membership model based on support vector data description (SVDD) is presented in this paper. The membership value to each input sample is confirmed according to its distance to the center of the smallest enclosing hypersphere in the feature space. The proposed model is applied to fuzzy support vector regression (FSVR) for 2-dimensional data set simulation and predicting the concentration of 4-carboxybenzaldhyde (4-CBA) in industrial purified terephthalic acid (PTA) oxidation process. Simulation results indicate that the proposed method actually reduces the error of regression and yields higher accuracy than support vector regression(SVR) does.