摘要
提出了一种基于GA的聚类集成算法ECUNGA(ensemble clustering using NMI and GA).算法利用GA搜索一个与聚类集体差异度小的聚类,以此来达到综合聚类集体信息,得到更优秀的聚类的目的.算法相比于传统基于互信息理论的方法,使用GA搜索,提高了搜索的能力且具有较低计算复杂度.最后,在UCI数据集上进行实验,取得了理想的效果.
A new clustering approach named ECUNGA was proposed based on GA and NMI.GA used to search for a better clustering,which had the lowest degree of difference with law clustering collective.ECUNGA had better performance on search optimal solution compared with traditional methods based on mutual information theory for the using of GA.It is well in the experiments on UCI data sets.
出处
《中国计量学院学报》
2011年第3期282-285,共4页
Journal of China Jiliang University
关键词
聚类
聚类集成
一致性函数
clustering
clustering ensemble
GA
consensus function