A hierarchical weighted low-rank representation for image clustering and classification
作者：
Highlights：
• High hierarchy neighbor is defined to capture more data points that may belong to the same subspace.
• A hierarchical weighted matrix can be obtained by probability to balance the contribution of the samples.
• A novel hierarchical weighted low rank representation is proposed to learn both local and global structure embedded in data better by affinity propagation.
• The representation matrix produced by HWLRR can work well with both clustering and classification methods.
摘要
•High hierarchy neighbor is defined to capture more data points that may belong to the same subspace.•A hierarchical weighted matrix can be obtained by probability to balance the contribution of the samples.•A novel hierarchical weighted low rank representation is proposed to learn both local and global structure embedded in data better by affinity propagation.•The representation matrix produced by HWLRR can work well with both clustering and classification methods.
论文关键词：Low-rank representation,Clustering,Semi-supervised learning,Similarity graph construction
论文评审过程：Received 23 May 2020, Revised 25 August 2020, Accepted 29 October 2020, Available online 4 November 2020, Version of Record 30 January 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107736
