A fuzzy-soft learning vector quantization

Abstract

This paper presents a batch competitive learning method called fuzzy-soft learning vector quantization (FSLVQ). The proposed FSLVQ is a batch type of clustering learning network by fusing the batch learning, soft competition and fuzzy membership functions. The comparisons between the well-known fuzzy LVQ and the proposed FSLVQ are made. In a series of designed simulations for the parameter estimations of normal mixtures, the performances including the accuracy (mean square error) and computational efficiency (number of iterations) are measured. FSLVQ shows good accuracy and high performance.