Weighted Proximal Support Vector Machines: Robust Classification

Despite of its great efficiency for pattern classification, proximal support vector machines (PSVM), a new version of SVM proposed recently, is sensitive to noise and outliers. To overcome the drawback, this paper modifies PSVM by associating a weight value with each input data of PSVM. The distance between each data point and the center of corresponding class is used to calculate the weight value. In this way, the effect of noise is reduced. The experiments indicate that new SVM, weighted proximal support vector machine (WPSVM), is much more robust to noise than PSVM without loss of computationally attractive feature of PSVM.