A Validity Index for Fuzzy and Possibilistic C-means Algorithm

Chunhui Zhang, Yiming Zhou, Trevor Martin.

This paper proposes a novel validity index for fuzzy-possibilistic c-means(FPCM) algorithm, it combines extended partition entropy and inter class similarity which is calculated from the fuzzy set point of view. The proposed index only requires the membership matrix and possibilistic(typicality) matrix, and is free from heavy distance computing. We also extend Xie-Beni index and Kwon index to evaluate FPCM. Experiments are done to compare the three indices and the results show its effectiveness.

PDF full paper