摘要
在聚类过程中利用先验信息能显著提高聚类算法的性能,但已存在的聚类融合算法很少考虑到数据集的先验信息。基于先验信息和谱分析,提出一种聚类融合算法,将成对限制信息引入到谱聚类算法中,用受限的谱聚类算法产生聚类成员,再采用基于互联合矩阵的集成方法生成最后的聚类结果。实验结果表明,利用先验信息能有效提高聚类的效果。
The prior knowledge can improve clustering performance, but few of clustering ensemble algorithms consider the prior knowledge of the datasets. This paper proposed a clustering ensemble algorithm based on prior knowledge and spectral analysis. It incorporated the pairwise constraints into the spectral clustering algorithm to generate clustering members. And obtained the final result by using the combining method of co-association matrix. The experimental results demonstrate that the proposed method can efficiently improve the clustering performance.
出处
《计算机应用研究》
CSCD
北大核心
2010年第6期2103-2105,共3页
Application Research of Computers
关键词
聚类融合
先验信息
成对限制
谱聚类
clustering ensemble
prior knowledge
pairwise constraints
spectral clustering
