A fuzzy C-means-based approach for selecting reference points in minimal learning machines

Abstract

This paper introduces a new approach to select reference points of minimal learning machines (MLM) for classification tasks. The proposal is based on the Fuzzy C-means algorithm and consists of selecting data samples from regions where no overlapping between classes exists. Such an idea has been empirically shown capable of achieving simpler decision boundaries in comparison to the standard MLM, and thus less susceptible to overfitting. Experiments were performed using UCI data sets. The proposal was able to both reduce the number of reference points and achieve competitive performance when compared to conventional approaches for selecting reference points.