Research map / Graph learning / GNN architectures
Permute Me Softly: Learning Soft Permutations for Graph Representations
Giannis Nikolentzos, George Dasoulas, Michalis Vazirgiannis
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 GIN message passing to produce node features instead of degree/triangle features (ablation) | pi-GNN · NCI1 | accuracy: 76.0 -> 80.4 | Improved |
| add GIN message passing to produce node features (ablation) | pi-GNN · REDDIT-BINARY | accuracy: 90.0 -> 92.0 | Improved |
| add GIN message passing to produce node features (ablation) | pi-GNN · REDDIT-MULTI-5K | accuracy: 53.2 -> 55.8 | Improved |
| add GIN message passing to produce node features (ablation) | pi-GNN · ENZYMES | accuracy: 60.3 -> 52.8 | Hurt |
| add GCN message passing to produce node features (ablation) | pi-GNN · COLLAB | accuracy: 73.1 -> 68.6 | Hurt |
| add GCN message passing to produce node features (ablation) | pi-GNN · IMDB-MULTI | accuracy: 48.5 -> 49.2 | Improved |
| replace structural features with MPNN node features (overall) (ablation) | pi-GNN · 10 graph classification datasets | accuracy | No clear effect |
| add dustbins (ablation) | pi-GNN · REDDIT-MULTI-5K | accuracy: 53.2 -> 49.1 | Hurt |
| add dustbins (ablation) | pi-GNN · REDDIT-BINARY | accuracy: 90.0 -> 87.9 | Hurt |
| add dustbins (ablation) | pi-GNN · COLLAB | accuracy: 73.1 -> 75.7 | Improved |
| add dustbins (ablation) | pi-GNN · ENZYMES | accuracy: 60.3 -> 56.8 | Hurt |
| add dustbins (ablation) | pi-GNN · ogbg-molhiv | ROC-AUC: 79.12 -> 79.09 | No clear effect |
Closest papers by meaning
- Going Deeper into Permutation-Sensitive Graph Neural Networks (ICML 2022 · similarity 0.92)
- SpeqNets: Sparsity-aware permutation-equivariant graph networks (ICML 2022 · similarity 0.90)
- Commute Graph Neural Networks (ICML 2025 · similarity 0.89)
- On the Expressivity and Sample Complexity of Node-Individualized Graph Neural Networks (NeurIPS 2024 · similarity 0.88)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.88)
- Covered Forest: Fine-grained generalization analysis of graph neural networks (ICML 2025 · 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.
