Semi-supervised learning is crucial in many applications where accessing class labels is unaffordable or costly. The most promising approaches are graph-based but they are transductive and they do not provide a generalized model working on inductive scenarios. To address this problem, we propose a generic framework, ESA☆, for inductive semi-supervised learning based on three components: an ensemble of semi-supervised autoencoders providing a new data representation that leverages the knowledge supplied by the reduced amount of available labels; a graph-based step that helps augmenting the training set with pseudo-labeled instances and, finally, a classifier trained with labeled and pseudo-labeled instances. Additionally, we also introduce two variants of our framework adopting different graph-based pseudo-labeling strategies: the first, ESALP, is based on a confidence-aware label propagation algorithm, while the second, ESAGAT, on a graph convolutional attention network. The experimental results show that our framework outperforms state-of-the-art inductive semi-supervised methods.

ESA*: A Generic Framework for Semi-supervised Inductive Learning

Shuyi Yang
First
;
Dino Ienco;Roberto Esposito;Ruggero G. Pensa
Last
2021

Abstract

Semi-supervised learning is crucial in many applications where accessing class labels is unaffordable or costly. The most promising approaches are graph-based but they are transductive and they do not provide a generalized model working on inductive scenarios. To address this problem, we propose a generic framework, ESA☆, for inductive semi-supervised learning based on three components: an ensemble of semi-supervised autoencoders providing a new data representation that leverages the knowledge supplied by the reduced amount of available labels; a graph-based step that helps augmenting the training set with pseudo-labeled instances and, finally, a classifier trained with labeled and pseudo-labeled instances. Additionally, we also introduce two variants of our framework adopting different graph-based pseudo-labeling strategies: the first, ESALP, is based on a confidence-aware label propagation algorithm, while the second, ESAGAT, on a graph convolutional attention network. The experimental results show that our framework outperforms state-of-the-art inductive semi-supervised methods.
447
102
117
https://www.sciencedirect.com/science/article/pii/S0925231221004252
semi-supervised learning, graph-based algorithms, inductive methods
Shuyi Yang, Dino Ienco, Roberto Esposito, Ruggero G. Pensa
File in questo prodotto:
File Dimensione Formato  
neucom2021_draft.pdf

Accesso aperto

Descrizione: PDF post-print
Tipo di file: POSTPRINT (VERSIONE FINALE DELL’AUTORE)
Dimensione 1.36 MB
Formato Adobe PDF
1.36 MB Adobe PDF Visualizza/Apri
neurocom2021_printed.pdf

Accesso riservato

Descrizione: PDF a stampa
Tipo di file: PDF EDITORIALE
Dimensione 2.03 MB
Formato Adobe PDF
2.03 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2318/1781701
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 0
social impact