Research map / Graph learning / GNN architectures
Rethinking Explaining Graph Neural Networks via Non-parametric Subgraph Matching
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 non-parametric subgraph matching to find explanatory subgraphs | MatchExplainer · MUTAG | ACC-AUC: 0.955±0.005 -> 0.997 | Helped |
| Use non-parametric subgraph matching to find explanatory subgraphs | MatchExplainer · VG-5 | ACC-AUC: 0.914±0.001 -> 0.993 | Helped |
| Use non-parametric subgraph matching to find explanatory subgraphs | MatchExplainer · MNIST | ACC-AUC: 0.636±0.003 -> 0.938 | Helped |
| Use non-parametric subgraph matching to find explanatory subgraphs | MatchExplainer · BA-3Motif | ACC-AUC: 0.586±0.004 -> 0.634 | Helped |
| Use non-parametric subgraph matching to find explanatory subgraphs | MatchExplainer · BA-3Motif | Recall@5: 0.297±0.000 -> 0.305 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GCN · MUTAG | testing accuracy (%): 0.828±0.004 -> 0.844±0.006 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GCN · VG-5 | testing accuracy (%): 0.619±0.003 -> 0.638±0.008 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GCN · MNIST | testing accuracy (%): 0.982±0.001 -> 0.986±0.002 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GIN · MUTAG | testing accuracy (%): 0.832±0.003 -> 0.845±0.007 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GIN · VG-5 | testing accuracy (%): 0.621±0.004 -> 0.630±0.003 | Helped |
| Preserve the explanatory subgraph and augment only the remaining graph | GIN · MNIST | testing accuracy (%): 0.988±0.001 -> 0.990±0.001 | Helped |
| Apply graph augmentation only to explanatory subgraphs (FPDrop) | GCN and GIN · MUTAG, VG-5, and MNIST | testing accuracy (%): 0.803±0.017; 0.587±0.014; 0.955±0.008; 0.806±0.020; 0.593±0.018; 0.959±0.005 | Hurt |
Closest papers by meaning
- EiG-Search: Generating Edge-Induced Subgraphs for GNN Explanation in Linear Time (ICML 2024 · similarity 0.90)
- Charting the Design Space of Neural Graph Representations for Subgraph Matching (ICLR 2025 · similarity 0.89)
- Generating In-Distribution Proxy Graphs for Explaining Graph Neural Networks (ICML 2024 · similarity 0.89)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.89)
- Iteratively Refined Early Interaction Alignment for Subgraph Matching based Graph Retrieval (NeurIPS 2024 · similarity 0.89)
- Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networks (ICLR 2024 · similarity 0.88)
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.