Research map / Graph learning / GNN architectures
Expressivity-Preserving GNN Simulation
Fabian Jogl, Maximilian Thiessen, Thomas Gärtner
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 |
|---|---|---|---|
| add DS graph transformation to GIN | GIN · CSL | accuracy: 0.1 ± 0.0 -> 1.0 ± 0.0 | Improved |
| add DSS graph transformation to GIN | GIN · CSL | accuracy: 0.1 ± 0.0 -> 1.0 ± 0.0 | Improved |
| add CWN graph transformation to GIN | GIN · CSL | accuracy: 0.1 ± 0.0 -> 1.0 ± 0.0 | Improved |
| DS transformation + GIN vs DS non-standard message passing | DS + GIN · 10 real-life datasets | fraction of datasets won: 60% | Improved |
| DS transformation + GCN vs DS non-standard message passing | DS + GCN · 10 real-life datasets | fraction of datasets won: 30% | Hurt |
| DSS transformation + GIN vs DSS non-standard message passing | DSS + GIN · 10 real-life datasets | fraction of datasets won: 60% | Improved |
| DSS transformation + GCN vs DSS non-standard message passing | DSS + GCN · 10 real-life datasets | fraction of datasets won: 30% | Hurt |
| CWN transformation + GIN vs CWN non-standard message passing | CWN + GIN · 10 real-life datasets | fraction of datasets won: 50% | No clear effect |
| CWN transformation + GCN vs CWN non-standard message passing | CWN + GCN · 10 real-life datasets | fraction of datasets won: 30% | Hurt |
| add graph transformations (DS, DSS) to MPNN baselines | GIN/GCN · 10 real-life datasets | fraction of datasets beating MPNN: 70% | Improved |
| add CWN transformation to GCN | GCN · 10 real-life datasets | fraction of datasets beating MPNN: 50% | No clear effect |
| use graph transformations with GIN vs GCN | DSS + GIN · 10 real-life datasets | fraction of datasets with best model: 30% | Improved |
Closest papers by meaning
- What Expressivity Theory Misses: Message Passing Complexity for GNNs (NeurIPS 2025 · similarity 0.89)
- Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness (ICLR 2024 · similarity 0.88)
- How Hard Is It for Message-Passing GNNs to Simulate One Weisfeiler-Lehman Color-Refinement Step? (NeurIPS 2026 · similarity 0.87)
- Towards Bridging Generalization and Expressivity of Graph Neural Networks (ICLR 2025 · similarity 0.87)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.86)
- An Empirical Study of Realized GNN Expressiveness (ICML 2024 · similarity 0.86)
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.
