基于稀疏一致图分解的鲁棒多视图聚类算法  

Robust multi-view clustering algorithm based on sparse consensus graph decomposition

作　　者：耿莉 王长鹏 GENG Li;WANG Changpeng(School of Science,Chang'an University,Xi'an 710064,China)

机构地区：[1]长安大学理学院,陕西西安710064

出　　处：《浙江大学学报（理学版）》2023年第5期569-579,共11页Journal of Zhejiang University（Science Edition）

基　　金：国家自然科学基金青年项目(12001057);长安大学中央高校基本科研业务费专项资金资助项目(300102122101);陕西省重点产业创新链项目(2020ZDLGY09-09);陕西省自然科学基础研究计划项目(2020JQ-346).

摘　　要：由于数据形式日益复杂,陆续涌现了大量多视图聚类算法。但现有方法存在计算复杂度较高、需要额外的后续处理步骤、构造的相似图非最优等缺点。基于此,首先提出一种基于稀疏一致图分解的单视图聚类算法,然后将其扩展为多视图聚类算法,考虑不同视图对最终结果的贡献不同,对每个视图分配适当的权重,同时利用L_(2.1)范数,得到性能更优的一致图,在一致图基础上学习非负表示矩阵,经交替迭代得到聚类结果。最后在多个数据集上进行比较实验,验证了该算法的有效性。Due to the increasing complexity of data form,multi-view clustering algorithms emerge one after another.The main disadvantages of existing methods include:the computational complexity of these methods is high;the final clustering involves additional processing steps;the similarity graph constructed may not be the optimal graph.In order to solve the above problems,a clustering algorithm based on sparse consensus graph decomposition is proposed.The algorithm is first tested on single-view data,and then extends from single-view data to multi-view data.The algorithm takes into account different contributions of different views to the final result by giving each view appropriate weight,at the same time,makes use of the L_(2,1)norm to obtain the consensus graph with better performance,learns the non-negative representation matrix on the basis of the consensus graph,and reveals the cluster result directly after alternation iteration.Finally,an update iterative algorithm is proposed and tested on a large number of data sets to verify the effectiveness of the algorithm.

关 键 词：多视图聚类 L_(2 1)范数 一致图分解 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]