A bi-level formulation for multiple kernel learning via self-paced training
作者：
Highlights：
• A novel optimization problem in a bi-level formulation for learning an affective multiple kernel.
• Capturing the accurate and complex relationships of data via global and local kernel alignments in a shared design.
• Evaluating the reliability of the training samples via self-paced learning.
• Robustness improvement of the proposed model via automatically absorbing the reliable instances into the model.
摘要
•A novel optimization problem in a bi-level formulation for learning an affective multiple kernel.•Capturing the accurate and complex relationships of data via global and local kernel alignments in a shared design.•Evaluating the reliability of the training samples via self-paced learning.•Robustness improvement of the proposed model via automatically absorbing the reliable instances into the model.
论文关键词：Multiple kernel learning,Self-paced learning,Bi-level optimization,Local kernel alignment,Global kernel alignment
论文评审过程：Received 21 April 2021, Revised 15 April 2022, Accepted 30 April 2022, Available online 2 May 2022, Version of Record 10 May 2022.
论文官网地址：https://doi.org/10.1016/j.patcog.2022.108770
