Multi-linear neighborhood preserving projection for face recognition
作者：
Highlights：
• A new method for supervised and unsupervised dimensionality reduction problem.
• Work in embedded manifold to allow representation of multi-dimensional images.
• Natural extension to 2-D, 3-D or higher-order tensors.
• Combine intrinsic geometry and discriminative information for face recognition.
• Avoid the small sample size and the curse of dimensionality problems.
摘要
•A new method for supervised and unsupervised dimensionality reduction problem.•Work in embedded manifold to allow representation of multi-dimensional images.•Natural extension to 2-D, 3-D or higher-order tensors.•Combine intrinsic geometry and discriminative information for face recognition.•Avoid the small sample size and the curse of dimensionality problems.
论文关键词：Feature extraction,Multi-linear projection,Neighborhood preserving projection (NPP),Tensor analysis,Face recognition,Dimensionality reduction
论文评审过程：Received 21 December 2012, Revised 5 July 2013, Accepted 1 August 2013, Available online 26 August 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.08.005
