GraphMix: Improved Training of GNNs for Semi-Supervised Learning

Vikas Verma*, Meng Qu, Kenji Kawaguchi, Alex Lamb, Yoshua Bengio, Juho Kannala, Jian Tang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

38 Citations (Scopus)

Abstract

We present GraphMix, a regularization method for Graph Neural Network based semi-supervised object classification, whereby we propose to train a fully-connected network jointly with the graph neural network via parameter sharing and interpolation-based regularization. Further, we provide a theoretical analysis of how GraphMix improves the generalization bounds of the underlying graph neural network, without making any assumptions about the "aggregation" layer or the depth of the graph neural networks. We experimentally validate this analysis by applying GraphMix to various architectures such as Graph Convolutional Networks, Graph Attention Networks and Graph-U-Net. Despite its simplicity, we demonstrate that GraphMix can consistently improve or closely match state-of-the-art performance using even simpler architectures such as Graph Convolutional Networks, across three established graph benchmarks: Cora, Citeseer and Pubmed citation network datasets, as well as three newly proposed datasets: Cora-Full, Co-author-CS and Co-author-Physics.

Original languageEnglish
Title of host publicationTHIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE
PublisherAAAI Press
Pages10024-10032
Number of pages9
Publication statusPublished - 2021
MoE publication typeA4 Conference publication
Event35th AAAI Conference on Artificial Intelligence / 33rd Conference on Innovative Applications of Artificial Intelligence / 11th Symposium on Educational Advances in Artificial Intelligence - Virtual, Online
Duration: 2 Feb 20219 Feb 2021

Publication series

NameAAAI Conference on Artificial Intelligence
PublisherAAAI
Volume35
ISSN (Print)2159-5399

Conference

Conference35th AAAI Conference on Artificial Intelligence / 33rd Conference on Innovative Applications of Artificial Intelligence / 11th Symposium on Educational Advances in Artificial Intelligence
CityVirtual, Online
Period02/02/202109/02/2021

Fingerprint

Dive into the research topics of 'GraphMix: Improved Training of GNNs for Semi-Supervised Learning'. Together they form a unique fingerprint.

Cite this