On Combining Nature-Inspired Algorithms for Data Clustering

Abstract

This chapter proposed different hybrid clustering methods based on combining particle swarm optimization (PSO), gravitational search algorithm (GSA) and free parameters central force optimization (CFO) with each other and with the k-means algorithm. The proposed methods were applied on 5 real datasets from the university of California, Irvine (UCI) machine learning repository. Comparative analysis was done in terms of three measures; the sum of intra cluster distances, the running time and the distances between the clusters centroids. The initial population for the used algorithms were enhanced to minimize the sum of intra cluster distances. Experimental results show that, increasing the number of iterations doesn't have a noticeable impact on the sum of intra cluster distances while it has a negative impact on the running time. K-means combined with GSA (KM-GSA), PSO combined with GSA (PSO-GSA) gave the best performance according to the sum of intra cluster distances while K-means combined with PSO (KM-PSO) and KM-GSA were the best in terms of the running time. Finally, KM-GSA and GSA have the best performance.

Cuckoo and krill herd‐based k‐means++ hybrid algorithms for clustering

“…In the past, various researchers have integrated different NIAs with k‐means algorithm with fairly efficient results (Ahmed, Shedeed, Hamad, & Tolba, ). These hybrid clustering algorithms improve quality of clusters using NIAs and thus have wide application.…”

A novel data clustering approach based on whale optimization algorithm

“…Moreover, the output produced by the k‐means algorithm strongly depends on the initial values of the cluster centers (Likas et al, 2003). To resolve these problems, several nature‐inspired algorithms have been proposed (Ahmed et al, 2017; Nanda & Panda, 2014). Chuang et al (2011) proposed an accelerated chaotic particle swarm optimization (PSO) for data clustering.…”