Managing Borderline and Noisy Examples in Imbalanced Classification by Combining SMOTE with Ensemble Filtering

Abstract

Imbalance data constitutes a great difficulty for most algorithms learning classifiers. However, as recent works claim, class imbalance is not a problem in itself and performance degradation is also associated with other factors related to the distribution of the data as the presence of noisy and borderline examples in the areas surrounding class boundaries. This contribution proposes to extend SMOTE with a noise filter called Iterative-Partitioning Filter (IPF), which can overcome these problems. The properties of this proposal are discussed in a controlled experimental study against SMOTE and its most well-known generalizations. The results show that the new proposal performs better than exiting SMOTE generalizations for all these different scenarios.