An improved bacterial colony optimization using opposition-based learning for data clustering

Data clustering is a technique for dividing data objects into groups based on their similarity. K-means is a simple, effective algorithm for clustering. But, k-means tends to converge to local optima and depends on the cluster’s initial values. To address the shortcomings of the k-means algorithm, many nature-inspired techniques have been used. This paper is offered an improved version of bacterial colony optimization (BCO) based on opposition-based learning (OBL) algorithm called OBL + BCO for data clustering. An OBL is used to increase the speed of the convergence rate and searching ability of BCO by computing the opposite solution to the present solution. The strength of the proposed data clustering technique is evaluated using several well-known UCI benchmark datasets. Different performance measures are considered to analyze the strength of the proposed OBL + BCO such as Rand index, Jaccard index, Beta index, Distance index, Objective values, and computational time. The experimental results demonstrated that the proposed OBL + BCO data clustering technique outperformed other data clustering techniques.