A Multi-Objective Genetic Algorithm with Fuzzy Relational Clustering for Automatic Data Clustering

A multi-objective genetic algorithm based fuzzy relational eigen vector centrality based clustering algorithm (FRECCA) called FRECCA-NSGA-II is proposed for automatic data clustering in this paper. A given dataset is spontaneously promoted into an optimal number of groups in a precise fuzzy partitions through the fuzzy clustering algorithm called FRECCA. This FRECCA algorithm operates on a similarity square matrix which is generated by comparing the pair wise similarities between data points. In most of the cases, fuzzy clustering methods cannot differentiate the geometric structures of clusters due to the cohesion and separation measures of fuzzy partition and using only centroid information for clustering. NSGA-II algorithm is therefore convenient to search for uniform fuzzy partitions for different cluster shapes. The two well-known cluster validity indices, cohesion and separation, are optimized concurrently through multi-objective NSGA-II. Binary encoding is used for chromosomes, here encode the centers as well as variable length numbers of clusters. Experimental results on benchmark data sets are given to demonstrate that the FRECCA-NSGA-II is capable of determining well-separated, hyperspherical and overlapping clusters. The superiority of the propose FRECCA-NSGA-II over the existing clustering algorithm is thoroughly elucidated for real-life benchmark data sets.