Supervised discriminative dimensionality reduction by learning multiple transformation operators
作者：
Highlights：
• Learning per-class transformations while enjoying a closed-form solution.
• Adding L2,1 norm to avoid overfitting while ensuring the transformation to be sparse.
• Learning discriminative features in the transformed space.
• Constructing the kernelized version of the proposed method.
• Providing a proof-convergence for the regularized version of the proposed method.
摘要
•Learning per-class transformations while enjoying a closed-form solution.•Adding L2,1 norm to avoid overfitting while ensuring the transformation to be sparse.•Learning discriminative features in the transformed space.•Constructing the kernelized version of the proposed method.•Providing a proof-convergence for the regularized version of the proposed method.
论文关键词：
论文评审过程：Received 2 June 2019, Revised 30 August 2020, Accepted 31 August 2020, Available online 3 September 2020, Version of Record 14 September 2020.
论文官网地址：https://doi.org/10.1016/j.eswa.2020.113958
