A modified intuitionistic fuzzy c-means algorithm incorporating hesitation degree

• Published in: Pattern recognition letters Vol. 122; pp. 45 - 52
• Main Authors: Verma, Hanuman, Gupta, Akshansh, Kumar, Dhirendra
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.05.2019; Elsevier Science Ltd
• Subjects: Algorithms; Brain; Clustering; Computer simulation; Fuzzy c-means; Fuzzy set theory; Fuzzy sets; Intuitionistic fuzzy c-means; Intuitionistic fuzzy set; Learning algorithms; Machine learning; Statistical tests; Intuitionistic fuzzy set; Clustering; Intuitionistic fuzzy c-means; Fuzzy c-means
• ISSN: 0167-8655; 1872-7344
• DOI: 10.1016/j.patrec.2019.02.017

•We proposed mIFCM algorithm to overcome the problem of the IFCM algorithms.•The mIFCM includes the hesitation degree in the centroids and membership degree.•The mIFCM has been tested on triangular data and simulated brain datasets.•A statistical test is performed to show the significant performance of algorithms. Fuzzy c-means (FCM) algorithm is an unsupervised machine learning algorithm and has been used in many applications. But, FCM does not consider hesitation in the case of imprecise data. The intuitionistic fuzzy c-means (IFCM) algorithm, which is based on intuitionistic fuzzy set theory, has been proposed in the literature to handle the hesitation during clustering. However, the IFCM still does not consider the hesitation properly. To overcome this problem of the IFCM, we proposed a modified intuitionistic fuzzy c-means (mIFCM) algorithm incorporating hesitation degree in this paper. We have generated the triangular dataset and tested the proposed mIFCM algorithm on the triangular dataset and also validated the algorithms on publicly available simulated brain data. The experimental results show that mIFCM performs better in comparison to existing intuitionistic fuzzy clustering algorithms. A nonparametric statistical test is also carried out to demonstrate the significant performance of the proposed mIFCM algorithm in comparison to other existing clustering algorithms.