Research map / Graph learning / GNN architectures
Path Integral Based Convolution and Pooling for Graph Neural Networks
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 |
|---|---|---|---|
| replace GCN/GIN convolution+pooling with PANConv+PANPool | graph classification GNN · PointPattern phi_RSA=0.4 | test accuracy (%): 85.1 -> 94.4 | Helped |
| replace GCNConv+TopKPool with PANConv+PANPool | graph classification GNN · AIDS | test accuracy (%): 79.25 -> 92.75 | Helped |
| replace SGConv+EdgePool with PANConv+PANPool | graph classification GNN · MUTAGEN | test accuracy (%): 70.05 -> 69.70 | Hurt |
| use PANConv+PANPool instead of GCNConv+SAGPool | graph classification GNN · PROTEINS_full | test accuracy (%): 72.65 -> 72.65 | No clear effect |
| use hybrid PANPool (Xp + beta diag(M)) instead of pooling on MX l2 norm (ablation) | PAN · AIDS | test accuracy (%): 80.6 -> 97.5 | Helped |
| use normalized MET diagonal (PANPool) instead of unnormalized in hybrid score (ablation) | PAN · PROTEINS | test accuracy (%): 69.7 -> 76.6 | Helped |
| use hybrid PANPool instead of unnormalized-MET-only pooling (PANUMPool) (ablation) | PAN · AIDS | test accuracy (%): 82.7 -> 97.5 | Helped |
| use hybrid PANPool instead of unnormalized hybrid pooling (PANXUMPool) (ablation) | PAN · PROTEINS_full | test accuracy (%): 72.6 -> 71.7 | Mixed |
| replace GCNConv+TopKPool with PANConv+PANPool | graph regression · QM7 | test MAE: 43.6 -> 42.8 | Helped |
| vary PANPool variant (UM, XUM, M, XHM) on QM7 (ablation) | PAN · QM7 | test MAE: 43.6 -> 43.0 | No clear effect |
Closest papers by meaning
- Geometry-Aware Edge Pooling for Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- Weisfeiler and Lehman Go Paths: Learning Topological Features via Path Complexes (AAAI 2024 · similarity 0.87)
- Improving Graph Neural Networks by Learning Continuous Edge Directions (ICLR 2025 · similarity 0.87)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.87)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 2025 · similarity 0.86)
- Non-convolutional graph neural networks. (NeurIPS 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.