Joint Learning of the Graph and the Data Representation for Graph-Based Semi-Supervised Learning
Mariana
Vargas-Vieyra
author
Aurélien
Bellet
author
Pascal
Denis
author
2020-12
text
Proceedings of the Graph-based Methods for Natural Language Processing (TextGraphs)
Dmitry
Ustalov
editor
Swapna
Somasundaran
editor
Alexander
Panchenko
editor
Fragkiskos
D
Malliaros
editor
Ioana
Hulpu\textcommabelows
editor
Peter
Jansen
editor
Abhik
Jana
editor
Association for Computational Linguistics
Barcelona, Spain (Online)
conference publication
Graph-based semi-supervised learning is appealing when labels are scarce but large amounts of unlabeled data are available. These methods typically use a heuristic strategy to construct the graph based on some fixed data representation, independently of the available labels. In this pa- per, we propose to jointly learn a data representation and a graph from both labeled and unlabeled data such that (i) the learned representation indirectly encodes the label information injected into the graph, and (ii) the graph provides a smooth topology with respect to the transformed data. Plugging the resulting graph and representation into existing graph-based semi-supervised learn- ing algorithms like label spreading and graph convolutional networks, we show that our approach outperforms standard graph construction methods on both synthetic data and real datasets.
vargas-vieyra-etal-2020-joint
10.18653/v1/2020.textgraphs-1.4
https://aclanthology.org/2020.textgraphs-1.4
2020-12
35
45