Simultaneous label inference and discriminant projection estimation through adaptive self-taught graphs
作者：
Highlights：
• A flexible and adaptive graph-based semi-supervised embedding is proposed.
• The soft labels and the linear transformation are simultaneously estimated.
• Smoothness of labels and projections is imposed using two hybrid graphs.
• The hybrid graph is set to an auto-weighed fusion of the data and label graphs.
• Performance is assessed on eight real image datasets.
摘要
•A flexible and adaptive graph-based semi-supervised embedding is proposed.•The soft labels and the linear transformation are simultaneously estimated.•Smoothness of labels and projections is imposed using two hybrid graphs.•The hybrid graph is set to an auto-weighed fusion of the data and label graphs.•Performance is assessed on eight real image datasets.
论文关键词：Auto-weighted graph fusion,Structured data,Semi-supervised learning,Soft labels,Discriminant embedding,Image categorization
论文评审过程：Received 1 February 2022, Revised 20 July 2022, Accepted 6 August 2022, Available online 12 August 2022, Version of Record 27 August 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.118480
