一种高效的全局K-均值算法

A Highly Effective Global K-means Algorithm

针对全局K-均值算法时间复杂度大的问题,提出一种增量选择初始聚类中心的新方法。选择数据集中周围分布最密集的样本作为第一个初始聚类中心,选择最小化目标函数贡献大,并且和已有聚类中心距离远的样本作为下一个初始聚类中心。改进算法减少了增量选取初始聚类中心时的计算量,降低了时间复杂度。实验证明,改进算法与全局K-均值算法、快速全局K-均值算法相比,在不影响聚类效果的基础上,减少了聚类时间,与优化初始聚类中心的算法相比,聚类效果更优。

Aiming at the problem of the global K-means algorithm time complexity,this paper proposes a new method about incremental selection of the initial clustering center. Selecting the data centralized and surrounded by the most in-tensive data sample as the first choice of the initial cluster center, selecting the minimum of the objective function＆#39;s contribution is large,in addition,it＆#39;s far away from the center of the existing cluster’s sample as the next initial clus-tering center. Improved algorithm to reduce the calculation about increment selecting the initial cluster center, to reduce the time complexity. Experiment proved that the improved algorithms compared with the global k-means algorithm and the fast global k-means algorithm,on the basis of don’t affect the clustering effect,reducing the clustering time,com-pared with the optimal initial clustering center algorithm,it’s better clustering effect.