Unsupervised feature selection for attributed graphs
作者：
Highlights：
• A novel method of unsupervised feature selection for attributed graphs is given.
• Correlation learned from both link & content are embedded into sparse learning.
• A regularization which exploits the group behavior of linked data is proposed.
• Algorithms for convex/nonconvex regularization cases of our model are designed.
• Extensive numerical experiments validate the advantage of our new method.
摘要
•A novel method of unsupervised feature selection for attributed graphs is given.•Correlation learned from both link & content are embedded into sparse learning.•A regularization which exploits the group behavior of linked data is proposed.•Algorithms for convex/nonconvex regularization cases of our model are designed.•Extensive numerical experiments validate the advantage of our new method.
论文关键词：Unsupervised feature selection,Attributed graphs,Sparse learning,ADMM,CCCP
论文评审过程：Received 19 June 2019, Revised 26 October 2020, Accepted 27 November 2020, Available online 28 November 2020, Version of Record 9 December 2020.
论文官网地址：https://doi.org/10.1016/j.eswa.2020.114402
