Research map / Graph learning / GNN architectures
PairNorm: Tackling Oversmoothing in GNNs
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 |
|---|---|---|---|
| Apply PairNorm to SGC | SGC · Cora SSNC-MV, 100% missing | accuracy: 0.423 -> 0.745 | Helped |
| Apply PairNorm to SGC | SGC · CiteSeer SSNC-MV, 100% missing | accuracy: 0.290 -> 0.486 | Helped |
| Apply PairNorm to SGC | SGC · PubMed SSNC-MV, 100% missing | accuracy: 0.399 -> 0.719 | Helped |
| Apply PairNorm to SGC | SGC · Cora SSNC-MV, 0% missing | accuracy: 0.815 -> 0.811 | No clear effect |
| Apply PairNorm-si to GCN | GCN · Cora SSNC-MV, 100% missing | accuracy: 0.582 -> 0.731 | Helped |
| Apply PairNorm-si to GCN | GCN · CiteSeer SSNC-MV, 100% missing | accuracy: 0.313 -> 0.498 | Helped |
| Apply PairNorm-si to GCN | GCN · PubMed SSNC-MV, 100% missing | accuracy: 0.449 -> 0.745 | Helped |
| Apply PairNorm-si to GCN | GCN · CiteSeer SSNC-MV, 0% missing | accuracy: 0.695 -> 0.660 | Hurt |
| Apply PairNorm-si to GAT | GAT · Cora SSNC-MV, 100% missing | accuracy: 0.653 -> 0.718 | Helped |
| Apply PairNorm-si to GAT | GAT · CiteSeer SSNC-MV, 100% missing | accuracy: 0.428 -> 0.483 | Helped |
| Apply PairNorm-si to GAT | GAT · PubMed SSNC-MV, 100% missing | accuracy: 0.631 -> 0.714 | Helped |
| Apply PairNorm-si to GAT | GAT · CiteSeer SSNC-MV, 0% missing | accuracy: 0.693 -> 0.670 | Hurt |
Closest papers by meaning
- Graph Neural Networks Do Not Always Oversmooth (NeurIPS 2024 · similarity 0.90)
- Residual Connections and Normalization Can Provably Prevent Oversmoothing in GNNs (ICLR 2025 · similarity 0.89)
- A Signed Graph Approach to Understanding and Mitigating Oversmoothing (NeurIPS 2025 · similarity 0.89)
- Deep Hypergraph Neural Networks with Tight Framelets (AAAI 2025 · similarity 0.88)
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning (ICLR 2025 · similarity 0.87)
- Locality-Aware Graph Rewiring in GNNs (ICLR 2024 · similarity 0.87)
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.