Research map / Graph learning / GNN architectures
Explainability in Graph Neural Networks: A Taxonomic Survey
Hao Yuan, Haiyang Yu, Shurui Gui, Shuiwang Ji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use subgraph-level explanation (SubgraphX) instead of edge masks (GNNExplainer) | explanation method · Graph-SST2 | Fidelity+ (prob): 0.0953 -> 0.3152 | Improved |
| use SubgraphX instead of random importance scores | explanation method · BA-2Motifs | Fidelity+ (prob): 0.4903 -> 0.8642 | Improved |
| use flow-based GNN-LRP instead of GNNExplainer | explanation method · Graph-Twitter | Fidelity+ (prob): 0.1452 -> 0.1931 | Improved |
| use GNNExplainer on node classification instead of graph classification | GNNExplainer · BA-Shapes vs Graph-SST2 | Fidelity+ (prob): 0.0953 -> 0.2925 | Mixed |
| use DeepLift (no graph-specific design) instead of random scores | DeepLift · BA-Shapes | Fidelity+ (prob): 0.1884 -> 0.0411 | Hurt |
| use gradient x input rule (GNN-GI) instead of random scores | GNN-GI · BA-2Motifs | Fidelity+ (prob): 0.4903 -> 0.0466 | Hurt |
| use SubgraphX instead of random importance scores | explanation method · Graph-Twitter | Fidelity- (prob): 0.2825 -> -0.1085 | Improved |
| use GNN-LRP instead of GNN-GI | explanation method · BA-shapes | explanation accuracy (AUC-ROC): 0.8369 -> 0.9243 | Improved |
| use GNNExplainer instead of PGExplainer | explanation method · BA-Community | explanation accuracy (AUC-ROC): 0.6843 -> 0.9194 | Improved |
| use DeepLift instead of GNN-LRP | explanation method · BA-shapes | explanation stability: 0.0432 vs 0.1872 | Mixed |
| raise the Sparsity threshold | soft-mask explanation methods · all datasets | Fidelity+ | Hurt |
Closest papers by meaning
- Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networks (ICLR 2024 · similarity 0.93)
- Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspective (ICML 2025 · similarity 0.91)
- Self-Interpretable Graph Learning with Sufficient and Necessary Explanations (AAAI 2024 · similarity 0.91)
- Generating In-Distribution Proxy Graphs for Explaining Graph Neural Networks (ICML 2024 · similarity 0.91)
- Factorized Explainer for Graph Neural Networks (AAAI 2024 · similarity 0.91)
- GNNX-BENCH: Unravelling the Utility of Perturbation-based GNN Explainers through In-depth Benchmarking (ICLR 2024 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
