Safety-aware Graph-based Semi-Supervised Learning

作者：

Highlights：

• We design a novel mechanism to develop safe GSSL.

• SaGSSL simultaneously realizes the graph selection and safe classifier learning.

• The OP in our algorithm can be solved by an alternating iterative method.

• SaGSSL achieves highly competitive performance compared to GSSL.

摘要

•We design a novel mechanism to develop safe GSSL.•SaGSSL simultaneously realizes the graph selection and safe classifier learning.•The OP in our algorithm can be solved by an alternating iterative method.•SaGSSL achieves highly competitive performance compared to GSSL.

论文关键词：Semi-supervised learning,Graph composite,Safety mechanism,Laplacian support vector machine

论文评审过程：Received 21 October 2017, Revised 8 March 2018, Accepted 25 April 2018, Available online 27 April 2018, Version of Record 3 May 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.04.031