TY - JOUR
T1 - Self-weighted spectral clustering with parameter-free constraint
AU - Zhang, Rui
AU - Nie, Feiping
AU - Li, Xuelong
N1 - Publisher Copyright:
© 2017 Elsevier B.V.
PY - 2017/6/7
Y1 - 2017/6/7
N2 - The constrained spectral clustering (or known as the semi-supervised spectral clustering) focuses on enhancing the clustering capability by utilizing the side information. In this paper, a novel constrained spectral clustering method is proposed based on deriving a sparse parameter-free similarity. Different from other works, the proposed method transforms the given pairwise constraints into the intrinsic graph similarity and the penalty graph similarity respectively instead of incorporating them into one single similarity. Besides, the optimal weight can be automatically achieved to balance the graph optimization problems between the intrinsic graph and the penalty graph. Equipped with a general framework of efficiently unraveling the bi-objective optimization, the proposed method could obtain both ratio cut and normalized cut clusterings via updating the weighted Laplacian matrix until convergence. Moreover, the proposed method is equivalent to the spectral clustering, when no side information is provided. Consequently, the effectiveness and the superiority of the proposed method are further verified both analytically and empirically.
AB - The constrained spectral clustering (or known as the semi-supervised spectral clustering) focuses on enhancing the clustering capability by utilizing the side information. In this paper, a novel constrained spectral clustering method is proposed based on deriving a sparse parameter-free similarity. Different from other works, the proposed method transforms the given pairwise constraints into the intrinsic graph similarity and the penalty graph similarity respectively instead of incorporating them into one single similarity. Besides, the optimal weight can be automatically achieved to balance the graph optimization problems between the intrinsic graph and the penalty graph. Equipped with a general framework of efficiently unraveling the bi-objective optimization, the proposed method could obtain both ratio cut and normalized cut clusterings via updating the weighted Laplacian matrix until convergence. Moreover, the proposed method is equivalent to the spectral clustering, when no side information is provided. Consequently, the effectiveness and the superiority of the proposed method are further verified both analytically and empirically.
KW - Constrained spectral clustering
KW - Parameter-free similarity
KW - Quadratic weighted optimization
UR - http://www.scopus.com/inward/record.url?scp=85013465385&partnerID=8YFLogxK
U2 - 10.1016/j.neucom.2017.01.085
DO - 10.1016/j.neucom.2017.01.085
M3 - 文章
AN - SCOPUS:85013465385
SN - 0925-2312
VL - 241
SP - 164
EP - 170
JO - Neurocomputing
JF - Neurocomputing
ER -