从多角度分析现有聚类算法(英文)

Analyzing Popular Clustering Algorithms from Different Viewpoints

聚类是数据挖掘中研究的重要问题之一.聚类分析就是把数据集分成簇,以使得簇内数据尽量相似,簇间数据尽量不同.不同的聚类方法采用不同的相似测度和技术.从以下3个角度分析现有流行聚类算法: (1)聚类尺度; (2)算法框架; (3)簇的表示.在此基础上,分析了一些综合或概括了一些其他方法的算法.由于分析从3个角度进行,所提出的方法能够涵盖,并区分绝大多数现有聚类算法.所做的工作是自调节聚类方法以及聚类基准测试研究的基础.

Clustering is widely studied in data mining community. It is used to partition data set into clusters so that intra-cluster data are similar and inter-cluster data are dissimilar. Different clustering methods use different similarity definition and techniques. Several popular clustering algorithms are analyzed from three different viewpoints: (1) clustering criteria, (2) cluster representation, and (3) algorithm framework. Furthermore, some new built algorithms, which mix or generalize some other algorithms, are introduced. Since the analysis is from several viewpoints, it can cover and distinguish most of the existing algorithms. It is the basis of the research of self-tuning algorithm and clustering benchmark.