Semi-supervised graph regularized nonnegative matrix factorization with local coordinate for image representation
作者：
Highlights：
• In our proposed SLNMF, which incorporates the label information and the local coordinate constraint into NMF as the additional constraints to enhance the performance of the clustering algorithm.
• SGLNMF reveals the intrinsic geometrical information of the data space by constructing the graph regularization, and also takes both the local coordinate constraints and the label information into account, so it can have better performance on the AC and NMI.
• We develop the corresponding iterative updating optimization schemes to derive the iterative updating rules of matric factors U and Z. More importantly, the convergence proof of our proposed SGLNMF is provided.
摘要
•In our proposed SLNMF, which incorporates the label information and the local coordinate constraint into NMF as the additional constraints to enhance the performance of the clustering algorithm.•SGLNMF reveals the intrinsic geometrical information of the data space by constructing the graph regularization, and also takes both the local coordinate constraints and the label information into account, so it can have better performance on the AC and NMI.•We develop the corresponding iterative updating optimization schemes to derive the iterative updating rules of matric factors U and Z. More importantly, the convergence proof of our proposed SGLNMF is provided.
论文关键词：Nonnegative matrix factorization,Graph regularization,Semi-supervised learning,Local coordinate
论文评审过程：Received 7 June 2021, Revised 21 September 2021, Accepted 28 November 2021, Available online 12 December 2021, Version of Record 22 December 2021.
论文官网地址：https://doi.org/10.1016/j.image.2021.116589
