Classification with Reject Option Using the Self-Organizing Map

Abstract. Reject option is a technique used to improve classifier’s reliability in decision support systems. It consists on withholding the automatic classification of an item, if the decision is considered not sufficiently reliable. The rejected item is then handled by a different classifier or by a human expert. The vast majority of the works on this issue have been concerned with implementing a reject option by endowing a supervised learning scheme (e.g., MLP, LVQ or SVM) with a re-ject mechanism. In this paper we introduce variants of the Self-Organizing Map (SOM), originally an unsupervised learning scheme, to act as supervised classi-fiers with reject option, and compare their performances with that of the MLP classifier