Research map / Graph learning / GNN architectures
Learning Parametrised Graph Shift Operators
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 |
|---|---|---|---|
| Learn PGSO parameters in GAT | GAT · Cora | accuracy: 83.0$\;\pm\;$0.7 -> 84.3$\;\pm\;$0.4 | Helped |
| Learn PGSO parameters in SGC | SGC · CiteSeer | accuracy: 72.5$\;\pm\;$0.6 -> 73.1$\;\pm\;$0.5 | Helped |
| Learn PGSO parameters in GCN | GCN · ogbn-arxiv | ROC-AUC: 71.82$\;\pm\;$0.22 -> 72.66$\;\pm\;$0.32 | Helped |
| Learn PGSO parameters in GIN | GIN · MUTAG | accuracy: 89.4$\;\pm\;$5.6 -> 91.9$\;\pm\;$3.1 | Helped |
| Learn layer-specific PGSO parameters | GCN · PTC-MR | accuracy: 64.4$\;\pm\;$1.9 -> 69.0$\;\pm\;$1.9 | Helped |
| Learn layer-specific PGSO parameters | GIN · IMDB-BINARY | accuracy: 75.1$\;\pm\;$5.1 -> 77.1$\;\pm\;$3.2 | Helped |
| Learn layer-specific PGSO parameters | GIN · IMDB-MULTI | accuracy: 52.3$\;\pm\;$2.8 -> 53.8$\;\pm\;$2.7 | Helped |
| Learn PGSO parameters in GCN | GCN · OGBG-MOLHIV | ROC-AUC: 76.06$\;\pm\;$0.97 -> 76.75$\;\pm\;$1.12 | Helped |
| Use a single shared PGSO parameter tuple instead of layer-specific tuples (ablation) | GNN-PGSO versus GNN-mPGSO · Real-world node and graph classification datasets | accuracy | Mixed |
| Initialize PGSO from a normalized Laplacian rather than the GCN operator, adjacency, or all-zeros (ablation) | GCN-PGSO on Cora · Cora | accuracy | Hurt |
Closest papers by meaning
- Neural Tangent Kernels Motivate Cross-Covariance Graphs in Neural Networks (ICML 2024 · similarity 0.82)
- GrokFormer: Graph Fourier Kolmogorov-Arnold Transformers (ICML 2025 · similarity 0.82)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.82)
- Graph neural networks and non-commuting operators (NeurIPS 2024 · similarity 0.82)
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning (ICLR 2025 · similarity 0.81)
- G-Adapter: Towards Structure-Aware Parameter-Efficient Transfer Learning for Graph Transformer Networks (AAAI 2024 · similarity 0.81)
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.