Data Augmentation for Graph Neural Networks

Zhao, Tong; Liu, Yozen; Neves, Lucio Pereira; Woodford, Oliver J.; Jiang, Meng; Shah, Neil

doi:10.48550/arxiv.2006.06830

Cited by 22 publications

(29 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, there is no theoretical guarantee that GDC can reduce the over-smoothing issue. Finally, similar to our work, [29] propose a two-step procedure for data augmentation in graph neural network. They firstly use graph auto-encoder (GAE) [12] to estimate the edge probability which is used for resampling in later procedure.…”

Section: A Graph Neural Networkmentioning

confidence: 78%

Training Graph Neural Networks by Graphon Estimation

Hu¹,

Fang²,

Lin³

2021

Preprint

View full text Add to dashboard Cite

In this work, we propose to train a graph neural network via resampling from a graphon estimate obtained from the underlying network data. More specifically, the graphon or the link probability matrix of the underlying network is first obtained from which a new network will be resampled and used during the training process at each layer. Due to the uncertainty induced from the resampling, it helps mitigate the well-known issue of over-smoothing in a graph neural network (GNN) model. Our framework is general, computationally efficient, and conceptually simple. Another appealing feature of our method is that it requires minimal additional tuning during the training process. Extensive numerical results show that our approach is competitive with and in many cases outperform the other over-smoothing reducing GNN training methods.

show abstract

Section: A Graph Neural Networkmentioning

confidence: 78%

Training Graph Neural Networks by Graphon Estimation

Hu¹,

Fang²,

Lin³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Most recently, various augmentation techniques for graphs have been introduced. e.g., node dropping (You et al 2020), edge modification (Jin et al 2021;Qiu et al 2020;Zhao et al 2020), subgraph extraction (Jiao et al 2020;Sun et al 2021b), attribute masking (Zhu et al 2020(Zhu et al , 2021 and others (Hassani and Khasahmadi 2020;Kefato and Girdzijauskas 2021;Suresh et al 2021). GRACE (Zhu et al 2020) randomly drops edges and masks node features to generate two augmented views of a graph.…”

Section: Related Workmentioning

confidence: 99%

Augmentation-Free Self-Supervised Learning on Graphs

Lee¹,

Lee²,

Park³

2021

Preprint

View full text Add to dashboard Cite

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 augmentationbased 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 AF-GRL. 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 realworld datasets demonstrate the superiority of AFGRL.

show abstract

“…Zhou et al (Zhou et al 2020) developed three heuristic methods to generate the virtual data and successfully yields an average improvement accuracy on graph classification tasks. In node classification tasks, Zhao et al (Zhao et al 2020) discovered that neural edge predictors can effectively encode class-homophilic structure to promote intra-class edges and demote inter-class edges in given graph structure and they leveraged these insights to improve performance in GNN-based node classification via edge prediction. Dong et al (Dong et al 2020) propagated the labels of the training set through the graph structure and expand the training set.…”

Section: Data Augmentationmentioning

confidence: 99%

Null Model-Based Data Augmentation for Graph Classification

Xuan¹,

Wang²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

In network science, null model is typically used to generate a series of graphs based on randomization under certain condition, which is widely used as a term of comparison to verify whether the networks in question display some non-trivial features, such as community structure. Since such non-trivial features may play a significant role in graph classification, the null model could provide a new perspective for regularization, so as to lead to the enhancement of classification performance. In this paper, we propose a novel data augmentation framework based on null model for graph classification, which contains four parts: feature ranking, graph data augmentation, data filtration, and model retraining. Moreover, in this framework, three heuristic null model generation methods are proposed for different features. Experiments are conducted on five famous benchmark datasets, and the results show that our framework has promising performance, providing a new direction of data augmentation for graph classification.

show abstract

Data Augmentation for Graph Neural Networks

Cited by 22 publications

References 0 publications

Training Graph Neural Networks by Graphon Estimation

Training Graph Neural Networks by Graphon Estimation

Augmentation-Free Self-Supervised Learning on Graphs

Null Model-Based Data Augmentation for Graph Classification

Contact Info

Product

Resources

About