Multiobjective Differential Crisp Clustering for Evaluation of Clusters Dynamically

Abstract

In this article, a crucial issue for finding the number of clusters dynamically is raised. Recent research shows that researchers have been devoted all of their time to investigate an algorithm for dynamic clustering on single objective criteria. However, multiobjective clustering for fixed number of clusters has an edge over the single objective clustering. This fact motivated us to present a new Dynamic Multiobjective Differential Crisp Clustering algorithm that encodes the cluster centers in its vectors and simultaneously optimizes the well-known DB index and CS measure for finding global compactness and separation among the clusters. In the final generation, it produces a set of non-dominated solutions, from which the best solution is selected by computing the Minkowski Score. The corresponding vector length provides the number of clusters. Results are demonstrated the effectiveness and superiority of the proposed algorithm both quantitatively and qualitatively.