Title
ESA☆: A generic framework for semi-supervised inductive learning
Abstract
•A new inductive framework for graph-based semi-supervised classification is proposed.•The proposed framework combines semi-supervised autoencoders and graph-based pseudo-labeling.•Two variants based on confidence-aware label propagation and graph attention networks are proposed.•The framework outperforms state-of-the-art competitors on data with very small amounts of labeled examples.•The framework is generic as it is designed to work on data of any kind.
Year
DOI
Venue
2021
10.1016/j.neucom.2021.03.051
Neurocomputing
Keywords
DocType
Volume
Semi-supervised learning,Graph-based algorithms,Inductive methods
Journal
447
ISSN
Citations 
PageRank 
0925-2312
0
0.34
References 
Authors
0
4
Name
Order
Citations
PageRank
Shuyi Yang100.34
Dino Ienco229542.01
Roberto Esposito36410.87
Ruggero G. Pensa435431.20