摘要
为改善单一聚类算法的聚类性能,提出一种基于自组织特征映射(SOM)的聚类集成算法。该算法利用多个具有差异性的聚类成员,将原始数据集转换成一个新的特征空间矩阵;然后计算各个聚类成员的聚类综合质量,并将其作为新特征空间矩阵的属性权重,最后利用SOM神经网络进行集成,产生最终的共识聚类结果。实验结果表明,与集成前的基聚类算法和其它聚类集成算法相比,该算法能够有效地提高聚类质量。
To improve the clustering performance ofa single clustering algorithm, a clustering ensemble algorithm based on self-organi- zing feature map is proposed. Firstly, the ordinary dataset is transformed into a new feature space matrix using different clustering solutions. Then the overall cluster quality is computed for each clustering solution as the weight of the attribute of the new feature space matrix. Finally, the consensus clustering result is generated by SOM neural network. The experimental results show that the proposed algorithm can effectively improve the clustering performance comparing with other clustering ensemble algorithms and the basis clustering algorithm before combination.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第22期4885-4888,共4页
Computer Engineering and Design
关键词
聚类集成
自组织特征映射
特征空间矩阵
聚类综合质量
属性权重
clustering ensemble
self-organizing feature map
feature space matrix
overall cluster quality
weight of the attribute
作者简介
作者简介:谭维(1985-),男,湖北荆州人,硕士研究生,研究方向为计算智能和数据挖掘;
杨燕(1964-),女,安徽合肥人,博士,教授,CCF会员,研究方向为计算智能、群体智能和数据挖掘。E-mail:tanwei1103@126.com