Improving ANNs performance on unbalanced data with an AUC-based learning algorithm

Abstract

This paper investigates the use of the Area Under the ROC Curve (AUC) as an alternative criteria for model selection in classification problems with unbalanced datasets. A novel algorithm, named here as AUCMLP, which incorporates AUC optimization into the Multi-layer Perceptron (MLPs) learning process is presented. The basic principle of AUCMLP is the solution of an optimization problem that aims at ranking quality as well as the separability of class distributions with respect to the threshold decision. Preliminary results achieved on real data, point out that our approach is promising, and can lead to better decision surfaces, specially under more severe unbalance conditions. © 2012 Springer-Verlag.