Low-rank tensor ring learning for multi-linear regression

作者：

Highlights：

• A generalized multi-linear regression is proposed based on low rank tensor ring decomposition.

• Two optimization models are built up for tensor ring ridge regression based on rank minimization and tensor factorization.

• Experiments on spatio-temporal forecasting tasks and 3D reconstruction of human motion demonstrate the enhanced performance.

摘要

•A generalized multi-linear regression is proposed based on low rank tensor ring decomposition.•Two optimization models are built up for tensor ring ridge regression based on rank minimization and tensor factorization.•Experiments on spatio-temporal forecasting tasks and 3D reconstruction of human motion demonstrate the enhanced performance.

论文关键词：Multilinear regression,Ridge regression,Tensor ring decomposition

论文评审过程：Received 19 March 2019, Revised 21 July 2020, Accepted 7 November 2020, Available online 7 November 2020, Version of Record 19 February 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107753