Normalized possibilistic clustering algorithms

A generalized approach to possibilistic clustering algorithms was proposed in [17], where the memberships are evaluated directly according to the data information using the fuzzy set theory, and the cluster centers are updated via a performance index. The computational experiments in[171 based on the generalized possibilistic clustering algorithms revealed that these clustering algorithms could not provide very stable results when clustering some data sets. Following that, a new performance index with possibility weights is used to obtain a new update equation for the cluster centers, which lead to the normalized possibilistic clustering algorithms. The performance and efficiency of a specific normalized possibilistic clustering algorithm are illustrated by numerical experiments based. A comparison with the generalized possibilistic clustering algorithms shows that the new algorithm could present very stable clustering results.