Opposition-based antlion optimizer using Cauchy distribution and its application to data clustering problem

Abstract

This paper proposes an improved version of antlion optimizer (ALO) to solve data clustering problem. In this work, Cauchy distribution-based random walk is employed in place of uniform distribution to jump out of local optima as a first strategy. Then opposition-based learning model is utilized in conjunction with acceleration coefficient to overcome the slow convergence of classical ALO as second strategy to propose opposition-based ALO using Cauchy distribution (OB-C-ALO). The performance of the proposed OB-C-ALO is evaluated over a set of benchmark problems of different varieties of characteristics and analysed statistically by performing Wilcoxon rank-sum test. The proposed version then utilizes K-means clustering by refining the clusters formed using K-means as objective function. The algorithm is evaluated on six data sets of UCI machine learning repository and compared with classical ALO and recently developed version of ALO, namely OB-L-ALO, over benchmark test problems as well as data clustering problem and proved to be better in terms of performance achieved. © 2019, Springer-Verlag London Ltd., part of Springer Nature.