基于特征加权和非负矩阵分解的多视角聚类算法

为了在多视角聚类过程中同时考虑特征权重和数据高维性问题,提出一种基于特征加权和非负矩阵分解的多视角聚类算法(Multiview Clustering Algorithm based on Feature Weighting and Non-negative Matrix Factorization,FWNMF-MC).FWNMF-MC算法根据每个视角中每个特征在聚类过程中的重要性,自动赋予不同的权值.通过将每个视角空间中的特征矩阵分解为基矩阵与系数矩阵的乘积,将多视角数据从高维空间映射到低维空间.为了有效利用每个视角信息挖掘聚簇结构,最大化每个视角在低维空间的一致性.最后实验结果表明FWNMF-MC算法的聚类效果明显优于已有的4种有代表性的多视角聚类算法.

A Multiview Clustering Algorithm Based on Feature Weighting and Non-negative Matrix Factorization

In order to simultaneously solve the two problems of feature weighting and high dimension in the process of multiview clustering,a multiview clustering algorithm based on feature weighting and non-negative matrix factorization ( FWNMF-MC) is proposed.According to the importance of each feature, FWNMF-MC automatically assigns different weights to different features.Multiview data is mapped from high dimensional space to low dimensional space by factorizing feature matrices into basis matrices and coefficient matrices.In order to take advantage of multiview information to mine cluster structure,the consensus of each view is maximized in low dimensional space.The experiment results show the perfor-mence of FWNMF-MC algorithm is superior to four existing classical multiview clustering algorithms.