Research map / Graph learning / GNN architectures
Task-Agnostic Graph Explanations
Yaochen Xie, Sumeet Katariya, Xianfeng Tang, Edward W Huang, Nikhil Rao, Karthik Subbian, 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 |
|---|---|---|---|
| task-agnostic TAGE vs task-specific PGExplainer trained on same task | TAGE · MoleculeNet BACE | fidelity: 0.252 -> 0.378 | Improved |
| TAGE vs PGExplainer on same task | TAGE · MoleculeNet HIV | fidelity: 0.473 -> 0.595 | Improved |
| TAGE vs PGExplainer on same task | TAGE · MoleculeNet BBBP | fidelity: 0.182 -> 0.193 | No clear effect |
| TAGE vs PGExplainer on same task | TAGE · PPI Task 2 | fidelity: 0.345 -> 0.499 | Improved |
| TAGE vs PGExplainer on same task | TAGE · E-commerce Sellers | fidelity: 0.3246 -> 0.6515 | Improved |
| evaluate PGExplainer on a task different from its training task | PGExplainer · MoleculeNet BACE | fidelity: 0.252 -> 0.007 | Hurt |
| evaluate PGExplainer on a task different from its training task | PGExplainer · MoleculeNet HIV | fidelity: 0.473 -> -0.001 | Hurt |
| train TAGE on individual MoleculeNet dataset instead of ZINC (ablation) | TAGE · MoleculeNet BACE | fidelity: 0.378 -> 0.402 | Mixed |
| train TAGE on individual MoleculeNet dataset instead of ZINC (ablation) | TAGE · MoleculeNet HIV | fidelity: 0.595 -> 0.541 | Mixed |
| TAGE trained on individual datasets vs PGExplainer (ablation) | TAGE · MoleculeNet SIDER | fidelity: 0.444 -> 0.516 | Improved |
| TAGE vs GEM causality-based explainer | TAGE · BACE sparsity 0.85 | fidelity: 0.3607 -> 0.4992 | Improved |
| TAGE vs PGExplainer at lower sparsity | TAGE · BACE sparsity 0.75 | fidelity: 0.5161 -> 0.5309 | No clear effect |
Closest papers by meaning
- 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 0.90)
- Factorized Explainer for Graph Neural Networks (AAAI 2024 · similarity 0.90)
- Stratified GNN Explanations through Sufficient Expansion (AAAI 2024 · similarity 0.89)
- Towards Robust Fidelity for Evaluating Explainability of Graph Neural Networks (ICLR 2024 · similarity 0.89)
- GraphTrail: Translating GNN Predictions into Human-Interpretable Logical Rules (NeurIPS 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.
