TY - JOUR
T1 - Multiview Fuzzy Clustering Based on Anchor Graph
AU - Yu, Weizhong
AU - Xing, Liyin
AU - Nie, Feiping
AU - Li, Xuelong
N1 - Publisher Copyright:
© 1993-2012 IEEE.
PY - 2024/3/1
Y1 - 2024/3/1
N2 - With the development of information technology, a large number of multiview data has emerged, which makes multiview clustering algorithms considerably attractive. Previous graph-based multiview clustering methods usually contain two steps: obtaining the fusion graph or spectral embedding of all views; and performing clustering algorithms. The two-step process cannot obtain optimal results since the two steps cannot negotiate with each other. To address this drawback, a novel algorithm named as multi-view fuzzy clustering based on anchor graph is presented. The proposed method can simultaneously obtain the membership matrix and minimize the disagreement rates of different views. A novel regularization based on trace norm is also presented in this article, which can not only obtain a clear clustering partition to prevent that all samples belonging to each cluster with the same membership value \frac{1}{c}, but also balance the size of each cluster. Moreover, we exploit the reweighted method to optimize the proposed model, which can introduce an adaptive weight to each view to deal with the unreliable views. A series of experiments are conducted on different datasets, and the clustering performance verifies the effectiveness and efficiency of the proposed algorithm.
AB - With the development of information technology, a large number of multiview data has emerged, which makes multiview clustering algorithms considerably attractive. Previous graph-based multiview clustering methods usually contain two steps: obtaining the fusion graph or spectral embedding of all views; and performing clustering algorithms. The two-step process cannot obtain optimal results since the two steps cannot negotiate with each other. To address this drawback, a novel algorithm named as multi-view fuzzy clustering based on anchor graph is presented. The proposed method can simultaneously obtain the membership matrix and minimize the disagreement rates of different views. A novel regularization based on trace norm is also presented in this article, which can not only obtain a clear clustering partition to prevent that all samples belonging to each cluster with the same membership value \frac{1}{c}, but also balance the size of each cluster. Moreover, we exploit the reweighted method to optimize the proposed model, which can introduce an adaptive weight to each view to deal with the unreliable views. A series of experiments are conducted on different datasets, and the clustering performance verifies the effectiveness and efficiency of the proposed algorithm.
KW - Fuzzy clustering
KW - graph
KW - multiview clustering
KW - reweighted optimization framework
UR - http://www.scopus.com/inward/record.url?scp=85168747216&partnerID=8YFLogxK
U2 - 10.1109/TFUZZ.2023.3306639
DO - 10.1109/TFUZZ.2023.3306639
M3 - 文章
AN - SCOPUS:85168747216
SN - 1063-6706
VL - 32
SP - 755
EP - 766
JO - IEEE Transactions on Fuzzy Systems
JF - IEEE Transactions on Fuzzy Systems
IS - 3
ER -