基于密度的加权K-Means算法

Weighted K-Means based on density

K-Means算法是比较流行的局域聚类算法,但由于其存在需要输入聚类数目以及对初始聚类中心敏感等缺陷,本文提出了一种基于密度的加权K-Means聚类算法来初始化聚类中心。该算法定义了点的密度函数和聚类中心函数,通过一定评价函数获取聚类中心。该方法获取的聚类中心不仅周围密度比较大,而且各个聚类中心之间相关性比较小,从而有效的减少了聚类时间,提高算法效率。

K-Means is one of the popular local area clustering algorithms. Aiming to its defects such as calling for the clustering number input and the sensitivity to initial clustering center, this paper put forward the advanced K-Means on the density and the weight to initialize the clustering center. The algorithm defines the point density function and the clustering center function, and obtains the clustering centers through the definition of the evaluation function. The clustering centers via the algorithm culd have not only high den- sity but also small relevance, and thereby reduce the time of clustering and improve the efficiency.