Exploring canonical correlation analysis with subspace and structured sparsity for web image annotation
作者：
Highlights：
• We model least-squares CCA with joint subspace extraction and structured sparsity.
• Low rank shared subspace is incorporated into CCA to capture label relationships.
• Structural sparse feature learning is employed for the canonical loadings.
• An efficient iterative algorithm is designed via the randomized technique.
• Evaluations show the effective Internet image annotation with high dimensionality.
摘要
•We model least-squares CCA with joint subspace extraction and structured sparsity.•Low rank shared subspace is incorporated into CCA to capture label relationships.•Structural sparse feature learning is employed for the canonical loadings.•An efficient iterative algorithm is designed via the randomized technique.•Evaluations show the effective Internet image annotation with high dimensionality.
论文关键词：Canonical correlation,Image annotation,Subspace learning,Sparsity
论文评审过程：Received 5 September 2014, Revised 2 September 2015, Accepted 25 June 2016, Available online 19 July 2016, Version of Record 6 August 2016.
论文官网地址：https://doi.org/10.1016/j.imavis.2016.06.008
