Classification method for imbalanced data based on spherical boundary
基于球边界的不平衡数据分类方法

Learning from data sets that contain very few instances of the positive class usually produces biased classifiers. They have a higher predictive accuracy over the negative class than that over the positive class (usually the more important class). A classification method for imbalance problem was proposed. The difference error penalties of two classes were introduced and the upper bounds of error rates could be controlled flexibly. The maximum separation ratio was obtained to separate two class instances with a single hypersphere, so the accuracies of classification and prediction over the positive class would be improved. Experiment results show that the method can effectively enhance the classification performance on imbalanced data sets.