Analysis of Membership Functions for Voronoi-Based Classification

Abstract

This paper addresses the problem of membership function selection for zoning-based classification. Different types of membership functions are considered based on abstract-level, ranked-level and measurement-level models and their effectiveness is estimated under different Voronoi-based zoning methods. The experimental tests, carried out in the field of hand-written numeral recognition, show that the best results are obtained when measurement-level models based on exponential models are used as membership functions.