An algorithm for automatic recognition of cluster centers based on local density clustering

Based on the local density clustering (LDC) algorithm, a new automatical local density clustering (ALDC) algorithm is proposed in this paper. Different from the existing LDC algorithm, the ALDC can capture the cluster center automatically. The new algorithm calculates the local density and the distance deviation of every point and expands the difference between the potential cluster center and other points by using these two features. The expansion of difference enables the machine automatically to recognize the cluster centers, and then assigns the remaining points by their closest neighbor of higher density. Experimental results on data sets show that the ALDC algorithm can achieve a accurate clustering and obtain a higher accuracy of clustering result than other two classical cluster algorithms.