Research map / Graph learning / GNN architectures
Global Explainability of GNNs via Logic Combination of Learned Concepts
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 |
|---|---|---|---|
| add straight-through discretization of concept activations (ablation) | GLGExplainer · Experiments across datasets | Fidelity | Helped |
| add straight-through discretization instead of using a concept-vector entropy loss (ablation) | GLGExplainer · Experiments across datasets | Fidelity | Helped |
| increase prototype count from 2 to 6 (ablation) | GLGExplainer · BAMultiShapes validation | Concept Purity: 0.42 -> 0.84 | Helped |
| increase prototype count from 6 to 8 (ablation) | GLGExplainer · BAMultiShapes validation | Fidelity: 0.95 -> 0.95 | No clear effect |
| increase prototype count from 2 to 8 (ablation) | GLGExplainer · Mutagenicity validation | Fidelity: 0.83 -> 0.79 | Hurt |
| increase prototype count from 2 to 4 (ablation) | GLGExplainer · HIN validation | Concept Purity: 0.45 -> 0.80 | Helped |
| increase prototype count from 4 to 8 (ablation) | GLGExplainer · HIN validation | Concept Purity: 0.80 -> 0.70 | Hurt |
Closest papers by meaning
- GraphTrail: Translating GNN Predictions into Human-Interpretable Logical Rules (NeurIPS 2024 · similarity 0.92)
- GNNX-BENCH: Unravelling the Utility of Perturbation-based GNN Explainers through In-depth Benchmarking (ICLR 2024 · similarity 0.91)
- GnnXemplar: Exemplars to Explanations - Natural Language Rules for Global GNN Interpretability (NeurIPS 2025 · similarity 0.91)
- Beyond Topological Self-Explainable GNNs: A Formal Explainability Perspective (ICML 2025 · similarity 0.91)
- Stratified GNN Explanations through Sufficient Expansion (AAAI 2024 · similarity 0.90)
- Self-Interpretable Graph Learning with Sufficient and Necessary Explanations (AAAI 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.