Research map / Graph learning / GNN architectures
Generative Causal Explanations for Graph Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Gem instead of GNNExplainer | GNN explanation · BA-shapes, K=5 | explanation accuracy: 82.4 -> 93.4 | Helped |
| Use Gem instead of the strongest baseline | GNN explanation · BA-shapes, K=6–9 | explanation accuracy: 88.2 -> 97.1; 91.2 -> 97.1; 91.2 -> 97.1; 94.1 -> 99.3 | Helped |
| Use Gem instead of PGExplainer | GNN explanation · Tree-cycles, K=6 | explanation accuracy: 94.4 -> 86.1 | Hurt |
| Use Gem instead of the strongest baseline | GNN explanation · Tree-cycles, K=7–9 | explanation accuracy: 80.6 -> 87.5; 74.6 -> 92.5; 91.4 -> 93.9 | Helped |
| Use Gem instead of GNNExplainer | GNN explanation · Tree-cycles, K=10 | explanation accuracy: 96.1 -> 95.4 | Hurt |
| Use Gem instead of GNNExplainer | GNN explanation · Mutag, K=15 | explanation accuracy: 67.1 -> 66.3 | Hurt |
| Use Gem instead of GNNExplainer | GNN explanation · Mutag, K=20–30 | explanation accuracy: 74.9 -> 78.0; 75.8 -> 82.1; 80.9 -> 83.4 | Helped |
| Use Gem instead of GNNExplainer | GNN explanation · NCI1, K=15 | explanation accuracy: 59.3 -> 56.9 | Hurt |
| Use Gem instead of GNNExplainer | GNN explanation · NCI1, K=20 | explanation accuracy: 61.8 -> 65.3 | Helped |
| Use Gem instead of GNNExplainer | GNN explanation · NCI1, K=25 | explanation accuracy: 69.6 -> 68.9 | Hurt |
| Use Gem instead of GNNExplainer | GNN explanation · NCI1, K=30 | explanation accuracy: 72.0 -> 72.8 | Helped |
| Amortize explanations with trained Gem rather than explaining instances individually | GNN explanation · BA-shapes | inference time per instance: 265.2 -> 0.5 ms | Helped |
Closest papers by meaning
- Self-Interpretable Graph Learning with Sufficient and Necessary Explanations (AAAI 2024 · similarity 0.91)
- Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspective (ICML 2025 · similarity 0.90)
- Factorized Explainer for Graph Neural Networks (AAAI 2024 · similarity 0.90)
- Stratified GNN Explanations through Sufficient Expansion (AAAI 2024 · similarity 0.89)
- GraphTrail: Translating GNN Predictions into Human-Interpretable Logical Rules (NeurIPS 2024 · similarity 0.89)
- Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networks (ICLR 2024 · similarity 0.89)
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.