A CLASSIFIER SYSTEM USING SMOOTH GRAPH COLORING

Abstract

Unsupervised classifiers allow clustering methods with less or no human intervention. Therefore it is desirable to group the set of items with less data processing. This paper proposes an unsupervised classifier system using the model of soft graph coloring. This method was tested with some classic instances in the literature and the results obtained were compared with classifications made with human intervention, yielding as good or better results than supervised classifiers, sometimes providing alternative classifications that considers additional information that humans did not considered.

Keywords

• coloración suave
• clasificación no supervisada
• clasificación automática
• agrupación
• optimización