基于Parzen窗的投影聚类方法

Projective Clustering Based on Parzen Window Technique

研究表明,高维数据的聚类都隐含在低维的子空间内,而这些子空间就是把原始数据投影到某些维度上的交集,于是相应的聚类算法就变成如何寻找合适的子空间内容。在此提出了一种新的划分子空间方法——基于Parzen窗子空间划分方法,并在这基础上提出了新的投影聚类方法PCPW。通过与最新的EPCH算法的实验结果对比表明,两者聚类效果相当,但PCPW算法更简单,易于实现。

Many researchers indicate that the clusters of a high dimensional dataset are often hidden in all the subspaces of the corresponding low dimensional datasets ,thus the corresponding clustering algorithm is to find appropriate subspaces. In this paper,a new Parzen-Window-based subspace dividing method is given,and accordingly,the new projective clustering algorithm PCPW is proposed. Compared with the latest EPCH algorithm,it has a comparable clustering performance,however,it can be more easily realized.