On Some Fuzzy Clustering Algorithms with Cluster-wise Covariance

Abstract

In many fuzzy clustering algorithms, KL-divergence-regularized method based on Gaus- sian mixture model (KFCM), Fuzzy Clasification Maximum Likelihood (FCML), and Fuzzy mixture of Student’s-t distributions (FSMM) have been proposed for cluster-wise covariate data, where various types of fuzzification technique have been applied to fuzzy clustering for cluster-wise isotropic data. In this report, some fuzzy clustering algorithms are proposed based on the combinations between five types of fuzzification including Bezdek-type fuzzification, KL-divergence regularization, Fuzzy Classifica- tion Maximum Likelihood, Tsallis-entropy regularization, and q-divergence-basis, and two types of mixture model including Gaussian mixture model and t-mixture model. Numerical experiments are conducted to show the features of the proposed methods.