An optimized initialization center K-means clustering algorithm based on density

Abstract

Traditional K-means algorithm's clustering effect is affected by the initial cluster center points. To solve this problem, a method is proposed to optimize the K-means initial center points. The algorithm use density-sensitive similarity measure to compute the density of objects. Through computing the minimum distance between the point and any other point with higher density, the candidate points are chosen out. Then, combined with the average density, the outliers are screened out. Ultimately the initial centers for K-means algorithm are screened out. Experimental results show that the algorithm gets the initial center points with high accuracy, and can effectively filter abnormal points. The running time and the iterations of the K-means algorithm are decreased obviously.