How Powerful are Graph Neural Networks?

Xu, Keyulu; Hu, Weihua; Leskovec, Jure; Jegelka, Stefanie

doi:10.48550/arxiv.1810.00826

Cited by 1,094 publications

(1,347 citation statements)

References 11 publications

Supporting

Mentioning

1,334

Contrasting

Unclassified

Order By: Relevance

“…GraphSAGE [19] aggregates neighborhood information via mean/max/LSTM pooling. GIN [45] allocates a learnable parameter for the center node when performing information aggregation, which empowers the model stronger capability to differentiate different graph structures. More recently, a myriad of more sophisticated aggregation strategies compared with previous works are developed and a more detailed review can be referred to [44,50].…”

Section: Resultsmentioning

confidence: 99%

“…Graph Neural Networks (GNNs) have demonstrated remarkable performance in a wide spectrum of graph learning tasks, e.g., node classification [19,24,45], link prediction [25,36,55], and recommendation [15,15,42]. The main intuition of GNNs is that they stack multiple layers of neural network primitives to learn high-level node feature representations, aiming at addressing various learning tasks in an end-to-end manner [13].…”

Section: Introductionmentioning

confidence: 99%

“…The seminal work of Graph Convolutional Network (GCN) [24] bridges the gap between these two families of algorithms. As a localized first-order approximation of spectral graph convolution [11], GCN can also be interpreted as a spatial-based method with a clear meaning of node localization, thus inspiring a lot of follow-up improvements [1,14,17,19,27,30,41,43,45,54], especially in the spatial domain. However, the fundamental studies and improvements of GCN from the spectral perspective is rather limited.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

AdaGNN

Dong

Ding

Jalaian

et al. 2021

Proceedings of the 30th ACM International Conference on Information &Amp; Knowledge Management

View full text Add to dashboard Cite

Graph Neural Networks have recently become a prevailing paradigm for various high-impact graph analytical problems. Existing efforts can be mainly categorized as spectral-based and spatialbased methods. The major challenge for the former is to find an appropriate graph filter to distill discriminative information from input signals for learning. Recently, myriads of explorations are made to achieve better graph filters, e.g., Graph Convolutional Network (GCN), which leverages Chebyshev polynomial truncation to seek an approximation of graph filters and bridge these two families of methods. Nevertheless, it has been shown in recent studies that GCN and its variants are essentially employing fixed low-pass filters to perform information denoising. Thus their learning capability is rather limited and may over-smooth node representations at deeper layers. To tackle these problems, we develop a novel graph neural network framework AdaGNN with a well-designed adaptive frequency response filter. At its core, AdaGNN leverages a simple but elegant trainable filter that spans across multiple layers to capture the varying importance of different frequency components for node representation learning. The inherent differences among different feature channels are also well captured by the filter. As such, it empowers AdaGNN with stronger expressiveness and naturally alleviates the over-smoothing problem. We empirically validate the effectiveness of the proposed framework on various benchmark datasets. Theoretical analysis is also provided to show the superiority of the proposed AdaGNN. The open-source implementation of AdaGNN can be found here: https://github.com/yushundong/AdaGNN.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AdaGNN

Dong

Ding

Jalaian

et al. 2021

Proceedings of the 30th ACM International Conference on Information &Amp; Knowledge Management

View full text Add to dashboard Cite

show abstract

“…We evaluate GStarX by explaining a standard GCN (Kipf & Welling, 2016) on all datasets in our major experiment. In later analysis, we also evaluate on GIN (Xu et al, 2018) and GAT (Veličković et al, 2017) on certain datasets following Yuan et al (2021). All models are trained to convergence.…”

Section: Experiments Settingsmentioning

confidence: 99%

Explaining Graph Neural Networks with Structure-Aware Cooperative Games

Zhang¹,

Liu²,

Shah³

et al. 2022

Preprint

View full text Add to dashboard Cite

Explaining predictions made by machine learning models is important and have attracted an increased interest. The Shapley value from cooperative game theory has been proposed as a prime approach to compute feature importances towards predictions, especially for images, text, tabular data, and recently graph neural networks (GNNs) on graphs. In this work, we revisit the appropriateness of the Shapley value for graph explanation, where the task is to identify the most important subgraph and constituent nodes for graph-level predictions. We purport that the Shapley value is a no-ideal choice for graph data because it is by definition not structure-aware. We propose a Graph Structure-aware eXplanation (GStarX) method to leverage the critical graph structure information to improve the explanation. Specifically, we propose a scoring function based on a new structure-aware value from the cooperative game theory called the HN value. When used to score node importance, the HN value utilizes graph structures to attribute cooperation surplus between neighbor nodes, resembling message passing in GNNs, so that node importance scores reflect not only the node feature importance, but also the structural roles. We demonstrate that GstarX produces qualitatively more intuitive explanations, and quantitatively improves over strong baselines on chemical graph property prediction and text graph sentiment classification.

show abstract

“…The entire model needs to be retrained if a new API call, which is not part of the training set, is encountered. Scott et al [22] proposed MalNet, a large scale Android malware FCG dataset, and they applied stateof-the-art graph representation learning approaches such as GIN [23] for Android malware classification. Among all the methods, Feather and GIN achieved the highest classification performance.…”

Section: Android Malware Detection Based On Graph Representation Lear...mentioning

confidence: 99%

Graph Neural Network-based Android Malware Classification with Jumping Knowledge

Lo¹,

Layeghy²,

Sarhan³

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper presents a new Android malware detection method based on Graph Neural Networks (GNNs) with Jumping-Knowledge (JK). Android function call graphs (FCGs) consist of a set of program functions and their interprocedural calls. Thus, this paper proposes a GNN-based method for Android malware detection by capturing meaningful intraprocedural call path patterns. In addition, a Jumping-Knowledge technique is applied to minimize the effect of the over-smoothing problem, which is common in GNNs. The proposed method has been extensively evaluated using two benchmark datasets. The results demonstrate the superiority of our approach compared to state-of-the-art approaches in terms of key classification metrics, which demonstrates the potential of GNNs in Android malware detection and classification.

show abstract

How Powerful are Graph Neural Networks?

Cited by 1,094 publications

References 11 publications

AdaGNN

AdaGNN

Explaining Graph Neural Networks with Structure-Aware Cooperative Games

Graph Neural Network-based Android Malware Classification with Jumping Knowledge

Contact Info

Product

Resources

About