基于自适应加权共识自表示的多视图子空间聚类  

Multi-view Subspace Clustering Based onAdaptive Weighted Consensus Self-representation

作　　者：李永 张维强[1] LI Yong;ZHANG Weiqiang(School of Mathematical Sciences,Shenzhen University,Shenzhen 518060,Guangdong Province,China)

机构地区：[1]深圳大学数学科学学院,广东深圳518060

出　　处：《吉林大学学报(理学版)》2025年第2期513-527,共15页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金(批准号:61972264);广东省自然科学基金(批准号:2019A1515010894);深圳市高校稳定支持计划项目(批准号:20200807165235002)。

摘　　要：针对如何充分融合多视图数据的互补性和多样性信息以提高聚类性能的问题,提出一种基于自适应加权共识自表示的多视图子空间聚类模型.首先,引入稀疏互斥性学习视图特定的稀疏自表示矩阵,再利用自适应加权学习多视图共识自表示矩阵以融合各视图所学到的自表示;其次,将多视图共识矩阵与聚类指示矩阵的学习整合到一个统一的优化模型,使自表示学习与聚类达到相互促进的效果;最后,在6个常用的多视图数据集上进行实验,并与9种相关方法进行对比.实验结果表明,该方法的信息融合效果明显,聚类效果有提升.Aiming at the problem of how to fully integrate the complementary and diverse information of multi-view data to improve the clust ering performance,we proposed a multi-view subspace clustering based on adaptive weighted consensus self-representation.Firstly,we introduced sparse mutual exclu sion to learn view-specific sparse self-representation matrix,and then used adaptive weighted learning of multi-view consensus self-representation matrix to fuse the self-representation learned from various views.Secondly,we integrated the learning of multi-view consensus matrix and clustering indicator matrix into a unified optimization model,so that self-representation learning and clustering could promote each other.Finally,we conducted experiments on six commonly used multi-view datasets,and compared them with nine related methods.The experimental results show that the proposed method has obvious information fusion effect and improves clustering effect.

关 键 词：多视图子空间聚类 稀疏表示 自表示 自适应加权学习 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]