Joint intermodal and intramodal correlation preservation for semi-paired learning
作者：
Highlights：
• A novel method is proposed to deal with learning in semi-paired scenarios.
• It is simple and effective to estimate the cross-view correlation relationship for unpaired data.
• By incorporating a rank constraint of Laplacian matrix, an ideal graph is obtained.
• The method jointly maximizes the cross-view correlation and within-view similarity simultaneously in the form of a generalized eigenvalue problem.
• Multi-view and non-linear cases also can be handled by the proposed learning method.
摘要
•A novel method is proposed to deal with learning in semi-paired scenarios.•It is simple and effective to estimate the cross-view correlation relationship for unpaired data.•By incorporating a rank constraint of Laplacian matrix, an ideal graph is obtained.•The method jointly maximizes the cross-view correlation and within-view similarity simultaneously in the form of a generalized eigenvalue problem.•Multi-view and non-linear cases also can be handled by the proposed learning method.
论文关键词：Semi-paired learning,Canonical correlation analysis,Clustering
论文评审过程：Received 16 July 2017, Revised 8 February 2018, Accepted 20 March 2018, Available online 27 March 2018, Version of Record 5 April 2018.
论文官网地址：https://doi.org/10.1016/j.patcog.2018.03.013
