Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Augmentation-Free Self-Supervised Learning on Graphs

About

Inspired by the recent success of self-supervised methods applied on images, self-supervised learning on graph structured data has seen rapid growth especially centered on augmentation-based contrastive methods. However, we argue that without carefully designed augmentation techniques, augmentations on graphs may behave arbitrarily in that the underlying semantics of graphs can drastically change. As a consequence, the performance of existing augmentation-based methods is highly dependent on the choice of augmentation scheme, i.e., hyperparameters associated with augmentations. In this paper, we propose a novel augmentation-free self-supervised learning framework for graphs, named AFGRL. Specifically, we generate an alternative view of a graph by discovering nodes that share the local structural information and the global semantics with the graph. Extensive experiments towards various node-level tasks, i.e., node classification, clustering, and similarity search on various real-world datasets demonstrate the superiority of AFGRL. The source code for AFGRL is available at https://github.com/Namkyeong/AFGRL.

Namkyeong Lee, Junseok Lee, Chanyoung Park• 2021

Related benchmarks

TaskDatasetResultRank
Node ClassificationCiteseer
Accuracy68.7
931
Node ClassificationPubmed
Accuracy79.71
819
Node ClassificationPhoto
Mean Accuracy93.25
343
Node ClassificationPubmed
Accuracy79.71
178
Molecular property predictionMoleculeNet BBBP (scaffold)
ROC AUC70.8
140
Node ClassificationPhoto
Accuracy93.25
139
Molecular property predictionMoleculeNet SIDER (scaffold)
ROC-AUC0.602
120
Molecular property predictionMoleculeNet BACE (scaffold)
ROC-AUC79.3
110
Molecular property predictionMoleculeNet MUV (scaffold)
ROC-AUC0.777
91
Node ClassificationComputer
Accuracy89.9
89
Showing 10 of 17 rows

Other info

Follow for update