Riemannian submanifold framework for log-Euclidean metric learning on symmetric positive definite manifolds

作者：

Highlights：

• We propose a Riemannian submanifold for log-Euclidean metric learning.

• We provide a simple analytic form of the derivative of a new distance function.

• We present several variants by modifying the proposed transformation matrices.

摘要

•We propose a Riemannian submanifold for log-Euclidean metric learning.•We provide a simple analytic form of the derivative of a new distance function.•We present several variants by modifying the proposed transformation matrices.

论文关键词：Riemannian submanifold,Log-Euclidean metric learning,Symmetric positive definite manifolds

论文评审过程：Received 11 January 2020, Revised 24 June 2021, Accepted 15 April 2022, Available online 27 April 2022, Version of Record 6 May 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117270