A novel clustering algorithm by adaptively merging sub-clusters based on the Normal-neighbor and Merging force

Junyi Guan, Sheng Li, Xiongxiong He, Jiajia Chen

Abstract: Clustering by fast search and find of density peaks (DPC) is a popular clustering method based on density and distance. In DPC, each non-center point’s cluster label is led by its nearest point with higher density, which may cause some misclassifications of non-center points and interfere with the choice of correct cluster centers in the decision graph. To avoid these defects, we propose a novel clustering algorithm that automatically generates clusters without using the decision graph based on the Normal-neighbor and Merging force (NM-DPC). We conduct a series of experiments on various challenging synthetic datasets. Experimental results demonstrate that NM-DPC can better identify clusters of complex shapes and automatically recognize the number of clusters.