GNNExplainer: Generating Explanations for Graph Neural Networks

By Rex Ying, Dylan Bourgeois, Jiaxuan You, Marinka Zitnik, Jure Leskovec. Neural Information Processing Systems, 2019

1,981 citations recorded by the source corpus.

Graph Neural Networks (GNNs) are a powerful tool for machine learning on graphs. GNNs combine node feature information with the graph structure by recursively passing neural messages along edges of the input graph. However, incorporating both graph structure and feature information leads to complex models and explaining predictions made by GNNs remains unsolved. Here we propose GnnExplainer, the first general, model-agnostic approach for providing interpretable explanations for predictions of any GNN-based model on any graph-based machine learning task. Given an instance, GnnExplainer identifies a compact subgraph structure and a small subset of node features that have a crucial role in GNN’s prediction. Further, GnnExplainer can generate consistent and concise explanations for an entire class of instances. We formulate GnnExplainer as an optimization task that maximizes the mutual information between a GNN’s prediction and distribution of possible subgraph structures. Experiments on synthetic and real-world graphs show that our approach can identify important graph structures as well as node features, and outperforms alternative baseline approaches by up to 43.0% in explanation accuracy. GnnExplainer provides a variety of benefits, from the ability to visualize semantically relevant structures to interpretability, to giving insights into errors of faulty GNNs.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/1510fe9ba6e9e576e46688b5c9f0f3950257175c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Xiangnan He
2. Xiang Wang
3. Tat-seng Chua
4. Jure Leskovec
5. Shuiwang Ji
6. Marinka Zitnik
7. Nicola De Cao
8. Pan Li

Citations in the corpus

• PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networks (2020, methodology): “We reuse the source code of [24] as we want to evaluate explainers on the same settings.”
• Discovery of a structural class of antibiotics with explainable deep learning (2023, positive): “We anticipate that future work will build on this and similar approaches 45,46 to further analyse and understand the predictions generated by deep learning models, for instance by using methods centred on perturbing model inputs 47 for additional tests of explainability, as well as perturbing…”
• Identity-aware Graph Neural Networks (2021, background): “…discovery (You et al. 2018a,b; Jin, Barzilay, and Jaakkola 2018), recommender systems (Ying et al. 2018; You et al. 2019a), and network analysis (Kumar, Cheng, and Leskovec 2017; Morris et al. 2019; Fan et al. 2019; Ying et al. 2019) stand to be greatly impacted by the use and development of GNNs.”
• Graph Neural Networks in Network Neuroscience (2021, methodology): “GNNExplainer and GraphLIME [77], [78] aimed to explain a graph by extracting the crucial features in the graph that affect the prediction of the GNN.”
• GNNGuard: Defending Graph Neural Networks against Adversarial Attacks (2020, background): “While deep learning on graphs has shown exciting results in a variety of applications [5, 21, 31, 32, 33], little attention has been paid to the robustness of such models, in contrast to an abundance of research for image (e.”
• Interpretable deep learning: interpretation, interpretability, trustworthiness, and beyond (2021, methodology): “We hope the taxonomy can shed light on future improvements/extensions on explaining Table 2 List of interpretation algorithm publications Methods Publications (non-exhaustive) LIME and variants LIME [137], Anchors [138], SHAP [110], RISE [127], MAPLE [130] Global interpretation LIME-SP [137], NormLI”
• GraphLIME: Local Interpretable Model Explanations for Graph Neural Networks (2020, methodology): “Recently, GNNexplainer [9] was proposed.”
• Explainable Deep Learning Models in Medical Image Analysis (2020, methodology): “or them is shown in1. 2.1. Model Specific vs Model Agnostic Model-specific interpretation methods are based on the parameters of the individual models. The graph neural network explainer (GNNExplainer) [11] is a special type of model-specific interpretability where the complexity of data representation”
• XGNN: Towards Model-Level Explanations of Graph Neural Networks (2020, methodology): “ing models on graph data become increasingly important but is still less explored. To the best of our knowledge, there is no existing study on interpreting GNNs at the model-level. The existing study [4, 40] only provides example-level explanations for graph models. As a radical departure from exist”
• From Artificial Intelligence to Explainable Artificial Intelligence in Industry 4.0: A Survey on What, How, and Where (2022, background): “, decision tree, SVM, random forest, linear regression, eXtreme gradient boosting (XGBoost) classification algorithm [39], graph neural network explainer [49] is a kind of this interpretability where the complex structure of data requires precise knowledge of the models.”
• Towards multi-modal causability with Graph Neural Networks enabling information fusion for explainable AI (2021, methodology): “Motivated by these problems, Ying et al. (2019) [64] proposed GNNExplainer, a model-agnostic approach that can provide interpretable explanations for predictions of any graph based model.”
• Explainability in Graph Neural Networks: A Taxonomic Survey (2020, methodology): “library, we include the implementations of several existing methods, such as GNNExplainer [46], PGExplainer [47], DeepLIFT [64], GNN-LRP [60], Grad-CAM [55], SubgraphX”