摘要： 在传统CLARANS聚类算法基础上，提出一种针对不确定性目标的CLARANS聚类算法。在该算法中，待聚类的每个不确定性目标都被表示成高斯混合模型，即高斯分布的一个加权和，并将Kullback-Leibler散度作为不确定性目标间的距离测度。在图片数据库上的实验结果表明，该算法具有较高的聚类精度。

关键词: 高斯分布, 高斯混合模型, Kullback-Leibler散度, CLARANS算法, 不确定性目标, 聚类算法

Abstract: Based on classical CLARANS clustering algorithm, a new clustering algorithm of uncertain objects is proposed in this paper. In the algorithm, each uncertain object is given as a Gaussian Mixture Model(GMM) which is the weighted sum of Gaussian distribution, and Kullback-Leibler Divergence(KLD) is used as distance measure between uncertain objects. Experimental result of image dataset shows the higher clustering precision of algorithm.

Key words: Gaussian distribution, Gaussian Mixture Model(GMM), Kullback-Leibler Divergence(KLD), CLARANS algorithm, uncertainty objects, clustering algorithm

中图分类号: 

• TP391.4