A unified weight learning and low-rank regression model for robust complex error modeling
作者：
Highlights：
• This is the first attempt to use a unified function to address both error distribution fitting and structure estimation.
• A generalized-correntropy-based weight learning theory is proposed to better fit complex error distribution in face images.
• A new low-rank approximation estimator based on the generalized correntropy is proposed for contiguous error structure estimation.
• We propose an optimization scheme based on the majorization minimization (MM) and alternating direction method of multipliers (ADMM).
摘要
•This is the first attempt to use a unified function to address both error distribution fitting and structure estimation.•A generalized-correntropy-based weight learning theory is proposed to better fit complex error distribution in face images.•A new low-rank approximation estimator based on the generalized correntropy is proposed for contiguous error structure estimation.•We propose an optimization scheme based on the majorization minimization (MM) and alternating direction method of multipliers (ADMM).
论文关键词：Regression,Weight learning,Low-rank approximation,Generalized correntropy,Robust learning
论文评审过程：Received 24 September 2020, Revised 16 April 2021, Accepted 30 June 2021, Available online 20 July 2021, Version of Record 27 July 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108147
