Using Unsupervised Learning to Guide Resampling in Imbalanced Data Sets

The class imbalance problem causes a classier to overt the data belonging to the class with the greatest number of training examples. The purpose of this paper is to argue that methods that equalize class membership are not as e ective as possible when applied blindly and that improvements can be obtained by adjusting for the within-class imbalance. A guided resampling technique is proposed and tested within a simpler letter recognition domain and a more diÆcult text classi cation domain. A fast unsupervised clustering technique, Principal Direction Divisive Partitioning (PDDP), is used to determine the internal characteristics of each class. The performance improvement in categories that su er from a large between-class imbalance (few positive examples) are shown to be improved when using the guided resampling method.