覆盖聚类算法

首先比较几类主要的聚类算法,给出每类算法的基本概念、原理、每类的代表性算法,及这些算法的主要特征。在此分析基础上,提出一种新的聚类算法———覆盖聚类算法,该算法采用覆盖的概念将比较集中的样本聚合在一起,从而发现隐含在样本集中的类,对于周围稀疏的样本结合最短距离法,获得聚类效果,并用实验数据对分层聚类方法、LBG方法与覆盖聚类算法进行比较,证明了覆盖聚类算法的可行性和有效性。最后给出了算法的研究方向。

Covering clustering algorithm

This paper firstly compares some main clustering algorithms. It presents the basic concepts, principles of each algorithm and representative clustering methods of each category algorithms, it compares the main characters of these algorithms. A new clustering algorithm,covering clustering algorithm is put forward in this paper. It uses the ideas of covering clustering the relatively dense samples,which are clusters.For the sparse samples, we use the method of the shortest distance interlinking them to achieve the results, we also compare the hierarchical clustering methods,LBG method and covering clustering method with experimental results. And the results show the feasibility and effectiveness of the covering clustering algorithm. Finally, this paper points out the research directions of the algorithm.