Rat Swarm Optimizer for Data Clustering

Abstract

Rat Swarm Optimization (RSO) is one of the newest swarm intelligence optimization algorithms that is inspired from the behaviors of chasing and fighting of rats in nature. In this paper we will apply the RSO to one of the most challenging problems, which is data clustering. The search capability of RSO is used here to find the best clusters centers. The proposed algorithm RSO for clustering (RSOC) is tested on several benchmarks and compared to some other optimization algorithms for data clustering including some wellknown and powerful algorithms such as Particle Swarm Optimization (PSO), and Genetic Algorithm (GA) and other recent such as the Hybridization of Krill Herd Algorithm and harmony search (H-KHA), hybrid Harris Hawks Optimization with differential evolution (H-HHO), and Multi-Verse Optimizer (MVO). Results are validated through a bunch of measures: homogeneity, completeness, v-measure, purity, and error rate. The computational results are encouraging, Where they demonstrate the effectiveness of RSOC over other techniques.