Geometric structure information based multi-objective function to increase fuzzy clustering performance with artificial and real-life data

Abstract

Grey wolf optimizer (GWO) is an efficient meta-heuristic algorithm that is inspired by the particular hunting behavior and leadership hierarchy of grey wolves in nature. In this paper, an efficient opposition-based grey wolf optimizer algorithm is proposed for solving the fuzzy clustering problem over artificial and real-life data. This work also tries to use the benefit of fuzzy properties which presents capability to handle overlapping clusters. However, centroid information and geometric structure information of clusters are the two important issues in fuzzy data clustering to improve the clustering performance. According to, in this paper, we derive two-objective functions, such as compactness and overlap–partition (OP) measures to handle above drawbacks. The centroid information issue is solved by compactness measure, and the OP measure is used to handle the geometric structure of clustering problem. Additionally, in the proposed clustering approach, the concept of opposition-based generation jumping and opposition-based population initialization is used with the standard GWO to enhance its computational speed and convergence profile. The efficiency of the proposed algorithm is shown for five artificial datasets and five real-life datasets of varying complexities. Experimental results show that the proposed method outperforms some existing methods with good clustering qualities.