Modified Kernel-based Intuitionistic Fuzzy C-means Clustering Method Using DNA Genetic Algorithm
Wenke Zang, Liyan Ren, Zhenni Jiang, Xiyu Liu · Journal of Software Engineering · 2017
Background: Clustering analysis has gained popularity and imprecise methods or their hybrid approaches has attracted many researchers of late.Fuzzy C-means clustering algorithm (FCM) is a method that is frequently used in pattern recognition.Recently, intuitionistic Fuzzy C-means (IFCM) algorithm was introduced and studied by Tripathy and it was found to be superior to all other algorithms in this family. Materials and Methods:This study proposes a modified IFCM method called kernel-based intuitionistic fuzzy C-means (mKIFCM) which is an extension of intuitionistic fuzzy C-means by adopting a kernel induced metric in the data space to replace the original Euclidean norm metric.The mKIFCM method combines Atanassovʼs Intuitionistic Fuzzy Entropy (IFE) with kernel-based fuzzy C-means and DNA genetic algorithms (DNA-GA) are optimally used simultaneously to choose the parameters of mKIFCM.The entire algorithm procedure is called mKIFCM-DNAGA.Results: The mKIFCM can make use of the advantages of intuitionistic fuzzy sets, kernel functions and DNA-GA in actual clustering problems.Conclusion: The algorithm is evaluated through cluster validity measures.The clustering accuracy of algorithm is investigated by classification datasets with labeled patterns.Experiments on machine learning repository datasets show that the proposed mKIFCM-DNAGA is more efficient than conventional algorithms.The mKIFCM-DNAGA method maintains appreciable performance compared to other methods in terms of pureness ratio.