Borderline over-sampling for imbalanced data classification

This paper proposes a new method for dealing with imbalanced data sets by over-sampling the borderline minority class instances by using a Support Vector Machine (SVM) classifier to predict future instances.