One-Sided Fuzzy SVM Based on Sphere for Imbalanced Data Sets Learning

Abstract

Learning from imbalanced data sets presents a new challenge to machine learning community, as traditional algorithms are biased to the majority classes and produce poor detection rate of the minority classes. This paper presents a one-sided fuzzy support vector machine algorithm based on sphere to improve the classification performance of the minority class. Firstly, the approach obtains the minimal hyper sphere of the majority class; secondly, it uses the center and radius of the hyper sphere to give the fuzzy membership of the majority instances, and thus effectively reduces the influence of majority noises and redundant instances in the classification process. Experiments show that our new approach improves not only the classification performance of the minority class more effectively, but also the classification performance of the whole data set comparing with other methods.