当前位置>主页 > 期刊在线 > 计算机技术 >

计算机技术22年15期

基于低秩约束的多视角聚类算法研究
贺艳芳,李莉杰
(河南开封科技传媒学院 理工学院,河南 开封 475004)

摘  要:随着信息技术获取的多样化,可以从不同角度和不同途径来获取不同的特征数据,如何有效挖掘多视角数据内部特征的一致性以及差异性是当前构造多视角聚类算法需要解决的问题。从视角的一致性出发,一方面引入核范数对多个视角之间的模糊隶属度矩阵进行低秩约束;另一方面研究多视角子空间聚类方法,利用低秩核映射,通过这种映射,使特征空间中的映射数据不仅具有低秩性,而且具有自表达性,从而使得低维子空间结构在高维特征空间中得以呈现。通过对比引入低秩约束多视角的算法,发现低秩约束的共性问题。


关键词:多视角数据;一致性;差异性;低秩约束



DOI:10.19850/j.cnki.2096-4706.2022.15.020


课题项目:河南省高等学校重点科研项目(21B520002);河南省高等学校重点科研项目(21B520003)


中图分类号:TP311.1                                       文献识别码:A                                文献编号:2096-4706(2022)15-0074-04


Research on Multi-view Clustering Algorithm Based on Low Rank Constraint

HE Yanfang, LI Lijie

(Institute of Technology, Henan Kaifeng College of Science Technology and Communication, Kaifeng 475004, China)

Abstract: With the diversification of information technology acquisition, different feature data can be obtained from different angles and different ways. How to effectively mine the consistency and difference of internal features of multi-view data is a problem to be solved in constructing multi-view clustering algorithm. Starting from consistency of perspectives, on the one hand, the kernel norm is introduced to carry out the low rank constraint of the fuzzy membership matrix between multiple perspectives. On the other hand, the multi view subspace clustering method is studied, and the low rank kernel mapping is used. Through this mapping, the mapping data in the feature space not only has low rank, but also has self-expression, so that the low dimensional subspace structure can be presented in the high dimensional feature space. By comparing the algorithms of introducing low rank constraint and multi view, it finds the common problems of low rank constraint.

Keywords: multi-view data; uniformity; difference; low rank constraint


参考文献:

[1] APPICE A,MALERBA D. A Co-Training Strategy for Multiple view Clustering in process Mining [J].IEEE Transactions on Services Computing,2016,9(6):832-845.

[2] ZHOU S H,ZHU E,LIU X W,et al. Subspace segmentation-based robust multiple kernel clustering [J].Information Fusion,2020,53:145-154.

[3] ZHAN K,ZHANG C Q,GUAN J P,et al. Graph learning for Multiview Clustering [J].IEEE Transactions on Cybernetics,2017,48(10):2887-2895.

[4] YIN Q Y,WU S,WANG L. Unified subspace learning for incomplete and unlabeled multi-view data [J].Pattern Recognition, 2017,67:313-327.

[5] EHSAN E,RENÉ V. Sparse Subspace Clustering: Algorithm,Theory,and Applications [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(11):2765-2781.

[6] LIU Y Y,JIAO L C,SHANG F H. An efficient matrix factorization based low-rank representation for subspace clustering [J]. Pattern Recognition,2013,46(1):284-292.

[7] KHEIRANDISHFARD M,ZOHRIZADEH F,KAMANGAR F. Deep low-rank subspace clustering [C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops(CVPRW). Seattle:IEEE,2020:3776-37781.

[8] 范莉莉,卢桂馥,唐肝翌,等 . 基于 Hessian 正则化和非负约束的低秩表示子空间聚类算法 [J]. 计算机应用,2022,42(1):115-122.

[9] 张涛,唐振民,吕建勇 . 一种基于低秩表示的子空间聚类改进算法 [J]. 电子与信息学报,2016,38(11):2811-2818.

[10] LEE B H,ABDULLAH J,K H A N Z A .Optimization of rapid prototyping parameters for production of flexible ABS object [J].Journal of materials processing technology,2005,169(1):54-61.

[11] MAHDI A,VISHAL M P. Deep multimodal subspace clustering networks [J].IEEE Journal of Selected Topics in Signal Processing,2018,12(6):1601-1614.

[12] 闫金涛,李钟毓,唐启凡,等 . 深度低秩多视角子空间聚类 [J]. 西安交通大学学报,2021,55(11):125-135.

[13] 张嘉旭,王骏,张春香,等 . 基于低秩约束的熵加权多视角模糊聚类算法 [J]. 自动化学报,2022,48(7):1760-1770.


作者简介:贺艳芳(1988—),女,汉族,河南漯河人,讲师,硕士研究生,主要研究方向:数据挖掘、人工智能、机器学习等;李莉杰(1988—),女,汉族,河南开封人,讲师,硕士研究生,主要研究方向:数据挖掘、大数据。