Tensor analysis with n-mode generalized difference subspace

作者：

Highlights：

• We propose a novel tensor data representation called n-mode GDS.

• The n-mode GDS is applied to the product manifold, providing a classification model.

• The n-mode GDS is optimized on the product manifold through a redefined Fisher score.

• We evaluate each tensor mode’s influence to develop a variant of the geodesic distance.

摘要

•We propose a novel tensor data representation called n-mode GDS.•The n-mode GDS is applied to the product manifold, providing a classification model.•The n-mode GDS is optimized on the product manifold through a redefined Fisher score.•We evaluate each tensor mode’s influence to develop a variant of the geodesic distance.

论文关键词：Action recognition,Tensor data classification,Generalized difference subspace,n-mode singular value decomposition

论文评审过程：Received 1 September 2019, Revised 18 November 2020, Accepted 30 December 2020, Available online 6 January 2021, Version of Record 19 January 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114559