Research map / Graph learning / GNN architectures
A Fractional Graph Laplacian Approach to Oversmoothing
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 |
|---|---|---|---|
| Use FLODE instead of strongest cited baseline | baseline GNN · Squirrel (undirected) | test accuracy: 59.22 -> 64.23 | Helped |
| Use FLODE instead of strongest cited baseline | baseline GNN · Chameleon (undirected) | test accuracy: 71.38 -> 73.60 | Helped |
| Use FLODE instead of strongest cited baseline | baseline GNN · Citeseer | test accuracy: 78.02 -> 78.07 | No clear effect |
| Use FLODE instead of strongest cited baseline | baseline GNN · Cora | test accuracy: 88.57 -> 86.44 | Hurt |
| Use FLODE instead of strongest cited baseline | baseline GNN · Pubmed | test accuracy: 90.23 -> 89.02 | Hurt |
| Use FLODE instead of strongest cited baseline | baseline GNN · Squirrel (directed) | test accuracy: 74.17 -> 74.03 | No clear effect |
| Use FLODE instead of strongest cited baseline | baseline GNN · Tolokers | AUROC: 83.78 -> 84.17 | Helped |
| Use FLODE instead of strongest cited baseline | baseline GNN · Roman-empire | accuracy: 88.75 -> 74.97 | Hurt |
| Use FLODE instead of strongest cited baseline | baseline GNN · DSBM alpha*=0.05 | accuracy: 94.1 -> 97.5 | Helped |
| Use FLODE instead of MLP | MLP · Squirrel | test accuracy: 28.77 -> 64.23 | Helped |
| Learnable fractional exponent alone instead of ODE framework with fixed alpha=1 (ablation) | FLODE · node classification | test accuracy | Helped |
| Remove learnable exponent, Schrodinger ODE or SNA directionality (ablation) | FLODE · Chameleon | test accuracy | Helped |
Closest papers by meaning
- Graph Neural Networks Do Not Always Oversmooth (NeurIPS 2024 · similarity 0.90)
- Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND (ICLR 2024 · similarity 0.90)
- A Dynamical Systems-Inspired Pruning Strategy for Addressing Oversmoothing in Graph Attention Networks (ICML 2025 · similarity 0.89)
- Distributed-Order Fractional Graph Operating Network (NeurIPS 2024 · similarity 0.89)
- Coupling Graph Neural Networks with Fractional Order Continuous Dynamics: A Robustness Study (AAAI 2024 · similarity 0.88)
- A Signed Graph Approach to Understanding and Mitigating Oversmoothing (NeurIPS 2025 · 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.