Generalized low-rank approximation of matrices based on multiple transformation pairs

作者：

Highlights：

• We proposed a novel multi-linear dimension reduction method.

• This use multi-pair of projection matrices in GLRAM method.

• The search space of this method is larger than GLRAM and less than SVD.

• At the same time have the benefits of GLRAM and SVD similtinously.

摘要

•We proposed a novel multi-linear dimension reduction method.•This use multi-pair of projection matrices in GLRAM method.•The search space of this method is larger than GLRAM and less than SVD.•At the same time have the benefits of GLRAM and SVD similtinously.

论文关键词：Machine learning,Matrix data classification,Kronecker product,Dimensionality reduction,SVD,GLRAM

论文评审过程：Received 6 February 2019, Revised 26 May 2020, Accepted 11 July 2020, Available online 14 July 2020, Version of Record 17 July 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107545