Embedding new observations via sparse-coding for non-linear manifold learning
作者：
Highlights：
• A framework based on sparse coding is proposed for out-of-sample embedding.
• The locality preserving property is jointly used with sparse coding.
• Classification performance and embedding consistency with batch modes are assessed.
• Experiments are conducted on six public face databases.
• K-nearest neighbor and Kernel Support Vector Machines classifiers are used.
摘要
Highlights•A framework based on sparse coding is proposed for out-of-sample embedding.•The locality preserving property is jointly used with sparse coding.•Classification performance and embedding consistency with batch modes are assessed.•Experiments are conducted on six public face databases.•K-nearest neighbor and Kernel Support Vector Machines classifiers are used.
论文关键词：Non-linear manifold learning,Out-of-sample embedding,Sparse representation,Face recognition
论文评审过程：Received 18 November 2012, Revised 18 May 2013, Accepted 11 June 2013, Available online 25 June 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.06.021
