Uma abordagem baseada em classificadores de larga margem para geração de dados artificiais em bases desbalanceadas

Abstract

In this work we propose the development of an approach capable of improving the results obtained by classification algorithms when applied to unbalanced datasets. The method, called Incremental Synthetic Balancing Algorithm (ISBA), performs an iterative procedure based on large margin classifiers, aiming to generate synthetic samples in order to reduce the level of unbalance. In the process, we use the support vectors as reference for the generation of new instances, allowing them to be positioned in regions with greater representativeness. Furthermore, the strategy allows the new samples to exceed the limits of the samples used as reference for their generation, which allows an extrapolation of the limits of the minority class, in order to achieve greater recognition of this class of interest. We present comparative experiments with other techniques, among them the Synthetic Minority Over-sampling Technique (SMOTE), which provide strong evidence of the applicability of the proposed approach.