Improved Cat Swarm Optimization to Solve Premature Convergence in Partitional Clustering
Feroz Ahmed, Sumit Kumar, Pradeep Kumar Singh · 2024
One of the vital activity in the area of data analysis is data clustering. It is an unsupervised learning used to make clusters of similar objects. The similarity between the objects are calculated by distance between the two clusters. Clustering algorithms are widely applied in solving real-time problems. The inter-cluster distance should be high and intra-cluster distance must be low for optimal clustering. The finding of optimal cluster is always a NP-hard problem. In this research, cat swarm optimization (CSO) is used to solve partitional clustering problem. CSO is a very efficient algorithm to solve the optimization problem but many times it stuck in premature convergence. Hence, an improved CSO is proposed to deal with these problems. The proposed algorithm is tested over some standard datasets and a comparative analysis is being done with some standard clustering algorithms. The results obtained reflects the supremacy of the proposed algorithm over existing clustering algorithms.