A convergence theorem and an experimental study of intuitionistic fuzzy c-mean algorithm over machine learning dataset

摘要

Clustering, which is one of the major unsupervised machine learning technique, plays an important role in many real world problems. For a clustering algorithm, the most desirable property is its guarantee to converge to a solution, which means mathematical convergence of the algorithm should be established. The convergence assures that clustering of a dataset always terminate after a finite number of iterations. So, in order to complete the study of intuitionistic fuzzy c-mean (IFCM) algorithm, which clusters intuitionistic fuzzy data, this paper gives the mathematical proof of its convergence. Also, the termination of the IFCM algorithm is experimentally demonstrated through its implementation over a number of real world datasets (UCI machine learning datasets) and synthetic datasets. The derived results clearly show the effectiveness of the intuitionistic fuzzy set based c-mean algorithms over FCM algorithm through standard measurement indexes. (C) 2018 Elsevier B.V. All rights reserved.