SNEOM: A Sanger Network Based Extended Over-Sampling Method. Application to Imbalanced Biomedical Datasets

摘要

In this work we introduce a novel over-sampling method to face the problem of imbalanced classes' classification. This method, based on the Sanger neural network, is capable of dealing with high-dimensional datasets. Moreover, it extends the capability of over-sampling methods and allows generating samples from both minority and majority classes. We have validated it in real medical applications where the involved datasets present an un-even representation among the classes and it has been obtained high sensitivities identifying minority classes. Therefore, by means of this method it is possible to accomplish the design of systems for the medical diagnosis with a high reliability.