Prototype based fuzzy clustering algorithms in high-dimensional feature spaces

Abstract

The ‘Curse of Dimensionality’ is formulated in a mathematicalway that is useful to understand the problems of clustering in high-dimensional spaces. Clustering tasks in high-dimensional spaces have a set of very difficult challenges, especially for one of the most widely used clustering algorithms: Fuzzy c-Means. Three alternatives to Fuzzy c-Means are described that can overcome its problems.