Locality preserving projection with symmetric graph embedding for unsupervised dimensionality reduction
作者：
Highlights：
• LPP_SGE is a new unsupervised projection and symmetric graph joint learning framework.
• LPP_SGE not only simultaneously considers the original space and subspace structures for graph learning but also considers the adaptive discriminative feature selection during feature extraction.
• With the proposed symmetric graph learning approach, it is possible to simultaneously acquire the Euclidean distance and linear representation relationships of samples in one term.
摘要
•LPP_SGE is a new unsupervised projection and symmetric graph joint learning framework.•LPP_SGE not only simultaneously considers the original space and subspace structures for graph learning but also considers the adaptive discriminative feature selection during feature extraction.•With the proposed symmetric graph learning approach, it is possible to simultaneously acquire the Euclidean distance and linear representation relationships of samples in one term.
论文关键词：Dimensionality reduction,Feature extraction,Graph embedding,Unsupervised learning
论文评审过程：Received 13 October 2021, Revised 29 May 2022, Accepted 9 June 2022, Available online 11 June 2022, Version of Record 14 June 2022.
论文官网地址：https://doi.org/10.1016/j.patcog.2022.108844
