利用CondEns算法实现聚类过程的改进

Implementing the Process of Clustering with CondEns Algorithm

在已知数据挖掘、聚类、数据聚类、非冗余等相关概念和基本聚类算法背景下,给出了非冗余聚类的总体概念框架和过程.引出了基于条件集非冗余聚类的改进聚类算法——CondEns算法,阐述了利用CondEns算法实现数集聚类的过程.最后在实验的基础上比较CondEns算法和没改进域聚类算法中的一种:CCIB算法,随着正交性减弱研究非冗余数集聚类的鲁棒性,发现了算法CondEns表现得比CCIB更好.

This paper presents a general concept framework and the process of non-redundant clustering under the background of knowing some concepts about data mining, clustering, data clustering, non-redundant and the fundamental clustering algorithm. The advanced clustering algorithm CondEns, which is based on the conditional ensembles non-redundant clustering, is elicited. Finally CondEns algorithm and one former domain clustering algorithm CCIB is compared by experiments. The results show that the robustness of non-redundant data ensemble clustering with the orthogonality assumption will be weakening, and the CondEns algorithm outperformed the CCIB algorithm.