Adaptive core fusion-based density peak clustering for complex data with arbitrary shapes and densities

Abstract A challenging issue of clustering in real-word application is to detect clusters with arbitrary shapes and densities in complex data. Many conventional clustering algorithms are capable of detecting non-spherical clusters, but their performance is limited when processing data with complex shapes and multiple density peaks in a cluster without knowing the number of clusters. This paper proposes an adaptive core fusion-based density peak clustering (CFDPC) for detecting clusters in any shape and density adaptively. An initial clustering based on automatic finding of density peaks is proposed first. An adaptive searching approach is then proposed to find core points, and a within-cluster similarity-based core fusion strategy is proposed to obtain the final clustering results. The CFDPC where the number of clusters arises intuitively is simple and efficient. The performance of CFDPC is successfully verified in clustering several benchmark complex datasets with diverse shapes and densities.