An imbalanced data classification algorithm based on boosting

It is currently a hot research topic that how to design appropriate learning algorithms to be applied to imbalanced data classification. This paper aims to investigate imbalanced data classification based on boosting and a weight-sampling boosting is proposed. The naïve loss function of boosting is modified by the sampling function, which makes the learned classifier focus on correct classification of positive samples. The experimental results performed on UCI data sets have shown that our algorithm outperforms naive boosting and previous algorithms in the problem of imbalanced data classification.