南京大学学报(自然科学版) ›› 2020, Vol. 56 ›› Issue (6): 862–869.doi: 10.13232/j.cnki.jnju.2020.06.008

• • 上一篇    

基于低秩稀疏约束的自权重多视角子空间聚类

夏菁, 丁世飞()   

  1. 中国矿业大学计算机科学与技术学院,徐州,221116
  • 收稿日期:2020-08-27 出版日期:2020-11-26 发布日期:2020-11-26
  • 通讯作者: 丁世飞 E-mail:dingsf@cumt.edu.cn
  • 作者简介:E⁃mail:dingsf@cumt.edu.cn
  • 基金资助:
    国家自然科学基金(61672522)

Self⁃weighted multi⁃view subspace clustering with low⁃rank sparse constrain

Jing Xia, Shifei Ding()   

  1. School of Computer Science and Technology,China University of Mining and Technology,Xuzhou,221116,China
  • Received:2020-08-27 Online:2020-11-26 Published:2020-11-26
  • Contact: Shifei Ding E-mail:dingsf@cumt.edu.cn

摘要:

多视角子空间聚类是一种利用视角之间的互补信息,找到视角间统一的表示并发现潜在分组结构的方法,近年来已成为机器学习的研究热点.提出一种基于低秩稀疏约束的自权重子空间聚类算法.具体的,低秩稀疏约束能发现数据的全局和局部结构信息,使自表示矩阵呈现稀疏性和低秩的特点;而自权重方法利用视角表示矩阵与共享相似度矩阵之间距离的反比为每个视角分配合理的权重,同时学习到一个视角之间共享的相似度矩阵,降低受损视角对于共享相似度矩阵的影响.以上提到的两种方法组成一个统一的优化框架,再使用增广拉格朗日乘子交换方向最小化方法(ALM?ADM)对提出的聚类算法进行优化.在基准数据集中的实验结果证明该算法比其他算法更有效.

关键词: 多视角子空间聚类, 自权重多视角聚类, 低秩稀疏约束, 多视角融合

Abstract:

Multi?view subspace clustering is a method which seeks a consensus representation among views and finds underlying grouping structure by using complementary information among views. In recent years,it has become a research hotspot in the field of machine learning. In this paper,we propose a self?weighted multi?view subspace clustering based on low?rank sparse constrain. Specifically,low?rank and sparse constraint can discover the global and local structural information of the data,making the self?representation matrix exhibit sparseness and low?rank characteristics. The self?weightinged method uses the inverse ratio of the distance between the representation matrix of views and the shared similarity matrix to assign proper weights to views,and the shared similarity matrix between the views is learned,which reduces the impact of the damaged view on the shared similarity matrix. The two mentioned above form a unified optimization framework. In this paper,the Augmented Lagrangian Multiplier with Alternating Direction Minimization(ALM?ADM) method is used to optimize the proposed clustering algorithm. Compared with other algorithms,experimental results in the benchmark dataset prove the effectiveness of our proposed algorithm.

Key words: multi?view subspace clustering, self?weight multi?view clustering, low?rank and sparse constrain, multi?view fusion

中图分类号: 

  • TP391

图1

Reuters数据集参数α1对算法性能的影响"

表1

Reuters数据集中算法性能的比较"

Reuters_1200ACCNMIPurity
SCrand40.76 (3.84)21.53 (2.60)59.29 (5.63)
LRSSCrand40.13 (2.80)24.50 (0.85)58.22 (3.31)
Co?reg20.62 (1.24)2.33 (0.34)20.95 (1.32)
DiMSC39.60 (1.32)18.17 (0.64)46.28 (1.74)
MVSC25.08 (0.39)6.60 (0.68)80.11 (5.50)
GFSC44.36 (2.20)25.49 (2.47)58.97 (1.90)
Ours140.68 (2.83)23.22 (1.81)57.80 (5.06)
Ours242.36 (4.29)23.37 (2.09)57.11 (4.32)
Ours45.53 (2.80)27.28 (1.16)59.80 (3.30)

表2

UCI digits数据集中算法性能的比较"

UCIACCNMIPurity
SCrand59.30 (4.08)57.35 (1.23)64.21 (1.24)
LRSSCrand78.12 (3.49)73.62 (1.98)78.22 (1.10)
Co?reg75.38 (7.35)71.97 (2.92)79.17 (4.98)
DiMSC42.72 (1.94)37.89 (0.87)45.65 (0.97)
MVSC79.60 (2.54)73.89 (1.93)87.19 (1.48)
GFSC78.97 (7.39)76.64 (3.05)83.68 (3.59)
Ours175.38 (5.63)74.93 (2.61)82.36 (3.20)
Ours279.17 (6.15)77.18 (4.01)80.31 (5.35)
Ours83.11 (4.62)81.20 (2.43)89.35 (3.69)

表3

3sources数据集中算法性能的比较"

3?sourcesACCNMIPurity
SCrand42.83 (1.02)43.66 (1.17)59.97 (1.24)
LRSSCrand47.29 (1.45)49.43 (1.98)60.22 (2.31)
Co?reg51.26 (3.17)50.23 (3.32)65.71 (3.50)
DiMSC69.23 (3.50)63.13 (3.25)75.74 (3.67)
MVSC57.43 (3.62)51.33 (2.21)67.26 (3.27)
GFSC65.37 (5.62)54.57 (5.29)69.96 (5.66)
Ours158.47 (1.37)50.5 (3.70)71.56 (1.54)
Ours266.95 (2.49)54.64 (4.25)75.68 (2.87)
Ours70.18 (2.67)57.39 (2.36)78.21 (3.66)
1 Chao G Q,Sun S L,Bi J B. A survey on multi?view clustering 2017,arXiv:1712.06246v2..
2 Wang Z Q,Yu Z W,Chen C L P,et al. Clustering by local gravitation. IEEE Transactions on Cybernetics,2018,48(5):1383-1396.
3 Liu G C,Lin Z C,Yu Y. Robust subspace segmentation by low?rank representation∥Proceedings of the 27th International Conference on International Conference on Machine Learning. Madison,WI,United States:ACM,2010:663-670.
4 Vidal R,Favaro P. Low rank subspace clustering (LRSC). Pattern Recognition Letters,2014,43:47-61.
5 Elhamifar E,Vidal R. Sparse subspace clustering:algorithm,theory and applications. IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(11):2765-2781.
6 Wang Y X,Xu H,Leng C L. Provable subspace clustering:when LRR meets SSC. IEEE Transactions on Information Theory,2019,65(9):5406-5432.
7 Li C G,Vidal R. Structured sparse subspace clustering:A unified optimization framework∥2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston,MA,USA:IEEE,2015:277-286.
8 Zhang J J,Li C G,Zhang H G,et al. Low?rank and structured sparse subspace clustering∥2016 Visual Communications and Image Processing. Chengdu,China:IEEE,2016:1-4.
9 Chen H Z,Wang W W,Feng X C. Structured sparse subspace clustering with within?cluster grouping. Pattern Recognition,2018,83:107-118.
10 Brbi? M,Kopriva I. Multi?view low?rank sparse subspace clustering. Pattern Recognition,2018,73:247-258.
11 Kumar A,Rai P,Daumé H. Co?regularized multi?view spectral clustering∥Proceedings of the 24th Neural Information Processing Systems. New York,NY,USA:Curran Associates Inc.,2011:1413-1421.
12 Cao X C,Zhang C Q,Fu H Z,et al. Diversity?induced multi?view subspace clustering∥2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston,MA,USA:IEEE,2015:586-594.
13 Wang X B,Guo X J,Lei Z,et al. Exclusivity?consistency regularized multi-view subspace clustering∥2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu,HI,USA:IEEE,2017:1-9.
14 Zhang C Q,Hu Q H,Fu H Z,et al. Latent multi?view subspace clustering∥2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu,HI,USA:IEEE,2017:4333-4341.
15 Zhang C Q,Fu H Z,Hu Q H,et al. Generalized latent multi?view subspace clustering. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(1):86-99.
16 Nie F P,Li J,Li X L. Self?weighted multiview clustering with multiple graphs∥Proceedings of the 26th International Joint Conference on Artificial Intelligence. Melbourne,Australia:AAAI Press,2017:2564-2570.
17 Kang Z,Shi G X,Huang S D,et al. Multi?graph fusion for multi?view spectral clustering. Knowledge?Based Systems,2020,189:105102.
18 Fan K. On a theorem of Weyl concerning eigenvalues of linear transformations:II. Proceedings of the National Academy of Sciences of the United States of America,1950,36(1):31-35.
19 Boyd S,Parikh N,Chu E,et al. Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends? in Machine Learning,2011,3(1):1-122.
20 Cai J F,Candès E J,Shen Z W. A singular value thresholding algorithm for matrix completion. SIAM Journal on Optimization,2010,20(4):1956-1982.
21 Donoho D L. De?noising by soft?thresholding. IEEE Transactions on Information Theory,1995,41(3):613-627.
22 Daubechies I,Defrise M,De Mol C. An iterative thresholding algorithm for linear inverse problems with a sparsity constraint. Communications on Pure and Applied Mathematics,2004,57(11):1413-1457.
23 Ng A Y,Jordan M I,Weiss Y. On spectral clustering:Analysis and an algorithm∥Proceedings of the 14th International Conference on Neural Information Processing Systems. Cambridge,MA,USA:MIT Press,2001:849-856.
24 Gao H C,Nie F P,Li X L,et al. Multi?view subspace clustering∥Proceedings of the 2015 IEEE International Conference on Computer Vision. Washington DC,USA:IEEE,2015:4238-4246.
[1] 朱伟,张帅,辛晓燕,李文飞,王骏,张建,王炜. 结合区域检测和注意力机制的胸片自动定位与识别[J]. 南京大学学报(自然科学版), 2020, 56(4): 591-600.
[2] 李昭阳,龚安民,伏云发. 基于EEG脑网络下肢动作视觉想象识别研究[J]. 南京大学学报(自然科学版), 2020, 56(4): 570-580.
[3] 郑建兴,李沁文,王素格,李德玉. 基于翻译模型的异质重边信息网络链路预测研究[J]. 南京大学学报(自然科学版), 2020, 56(4): 541-548.
[4] 黄雨婷,徐媛媛,张恒汝,闵帆. 融合标签结构依赖性的标签分布学习[J]. 南京大学学报(自然科学版), 2020, 56(4): 524-532.
[5] 任睿,张超,庞继芳. 有限理性下多粒度q⁃RO模糊粗糙集的最优粒度选择及其在并购对象选择中的应用[J]. 南京大学学报(自然科学版), 2020, 56(4): 452-460.
[6] 陈俊芬,赵佳成,韩洁,翟俊海. 基于深度特征表示的Softmax聚类算法[J]. 南京大学学报(自然科学版), 2020, 56(4): 533-540.
[7] 王宝丽,姚一豫. 信息表中约简补集对及其一般定义[J]. 南京大学学报(自然科学版), 2020, 56(4): 461-468.
[8] 陈石,张兴敢. 基于小波包能量熵和随机森林的级联H桥多电平逆变器故障诊断[J]. 南京大学学报(自然科学版), 2020, 56(2): 284-289.
[9] 周昊,沈庆宏. 基于改进音形码的中文敏感词检测算法[J]. 南京大学学报(自然科学版), 2020, 56(2): 270-277.
[10] 罗春春,郝晓燕. 基于双重注意力模型的微博情感倾向性分析[J]. 南京大学学报(自然科学版), 2020, 56(2): 236-243.
[11] 王露,王士同. 改进模糊聚类在医疗卫生数据的Takagi⁃Sugeno模糊模型[J]. 南京大学学报(自然科学版), 2020, 56(2): 186-196.
[12] 陈睿, 伏云发. 基于EEG握力变化及想象单次识别研究[J]. 南京大学学报(自然科学版), 2020, 56(2): 159-166.
[13] 俞冬明, 李苑, 李智星, 王国胤. 一种基于用户结构和属性的无监督用户对齐方法[J]. 南京大学学报(自然科学版), 2020, 56(1): 1-8.
[14] 王吉地,郭军军,黄于欣,高盛祥,余正涛,张亚飞. 融合依存信息和卷积神经网络的越南语新闻事件检测[J]. 南京大学学报(自然科学版), 2020, 56(1): 125-131.
[15] 陈超逸,林耀进,唐莉,王晨曦. 基于邻域交互增益信息的多标记流特征选择算法[J]. 南京大学学报(自然科学版), 2020, 56(1): 30-40.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!