Research map / Graph learning / GNN architectures
On Explainability of Graph Neural Networks via Subgraph Explorations
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 |
|---|---|---|---|
| identify connected subgraphs directly and score them with Shapley values using MCTS | SubgraphX explanations of graph classification GNNs · Five of six experiments across the evaluated datasets and GNNs | Fidelity | Helped |
| identify connected subgraphs directly and score them with Shapley values using MCTS | SubgraphX explanations on BA-Shape (GCN) · BA-Shape | Fidelity | Mixed |
| approximate Shapley values by restricting sampled players to the L-hop neighborhood (ablation) | SubgraphX on BBBP · BBBP (50 graphs, average 24.96 nodes) | time: 865.4±1.6s -> 77.8±3.8s | Helped |
| approximate Shapley values by restricting sampled players to the L-hop neighborhood (ablation) | SubgraphX on BBBP · BBBP (50 graphs, average 24.96 nodes) | Fidelity: 0.53 -> 0.55 | No clear effect |
| use Monte Carlo sampling rather than enumerating all Shapley coalitions (ablation) | SubgraphX on BBBP · BBBP (50 graphs, average 24.96 nodes) | time: >10 hours -> 77.8±3.8s | Helped |
| use Low2high rather than High2low pruning actions (ablation) | SubgraphX on BBBP · BBBP (50 graphs, average 24.96 nodes; explanations under 15 nodes) | Fidelity: 0.61046 -> 0.66149 | Helped |
| use High2low rather than Low2high pruning actions (ablation) | SubgraphX on BBBP · BBBP (50 graphs, average 24.96 nodes; explanations under 15 nodes) | time: 107.24s -> 21.52s | Helped |
Closest papers by meaning
- EiG-Search: Generating Edge-Induced Subgraphs for GNN Explanation in Linear Time (ICML 2024 · similarity 0.93)
- Generating In-Distribution Proxy Graphs for Explaining Graph Neural Networks (ICML 2024 · similarity 0.92)
- Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networks (ICLR 2024 · similarity 0.90)
- Graph Segmentation and Contrastive Enhanced Explainer for Graph Neural Networks (AAAI 2025 · similarity 0.90)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.90)
- GOAt: Explaining Graph Neural Networks via Graph Output Attribution (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.