一种改进的基于密度和网格的高维聚类算法  

Validity Validation of An Improved High-dimensional Cluster Analysis Algorithm Based on Grid and Intensity

作　　者：朱倩[1] 黄志军[1] 

机构地区：[1]海军工程大学,武汉430033

出　　处：《舰船电子工程》2005年第5期55-56,59,共3页Ship Electronic Engineering

摘　　要：提出了一种改进的基于密度和网格的高维聚类算法,并对算法有效性进行了验证。该算法通过减少样本点数量的方法达到减少稠密子空间数量。在发现高维稠密子空间时,对样本库进行精简。这些样本点的求得能有效减少求解最小聚类的时间复杂度。This paper proposes an improved high- dinaensional cluster analysis algorithm based on grid and intensity , then discusses it＇s validity validation. The amount of the density subspace can be deduced by cutting down that of sample data . The sampie library is simplified as the high - dimensional subspaces are found. By working out such sample data the time complexity of figuring out min cluster is effectively reduced.

关 键 词：数据挖掘 聚类 网格 密度 高维数据 子空间 最小聚类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]