A hybrid strategy for imbalanced classification

This paper describes a new hybrid strategy for highly imbalanced classification. Firstly we devise an adaptive scheme for minority generating; secondly, with data cleaning majority new clusters are drawn to increasingly focus on the combination of new minority samples. Inspired by the essence of SVM, our approach extracts the most informative SVs to train. An empirical study compares the performance of our approach with that of traditional classification approaches on the benchmark data sets. We evaluate the new hybrid strategy on 6 datasets from the UCI repository, and experimental results demonstrate the hybrid strategy not only inherent data distribution, but also improve classification effectiveness and accuracy.