Research map / Graph learning / GNN architectures
Weisfeiler and Lehman Go Cellular: CW 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 |
|---|---|---|---|
| Use induced-cycle 2-cells and hierarchical cellular message passing | CIN · CSL | accuracy: 10.000±0.000 -> 100.000±0.000 | Helped |
| Use induced-cycle cells for molecular graph message passing | CIN · ZINC, no edge features | MAE: 0.139±0.007 -> 0.115±0.003 | Helped |
| Use induced-cycle cells for molecular graph message passing | CIN · ZINC, with edge features | MAE: 0.108±0.018 -> 0.079±0.006 | Helped |
| Use induced-cycle cells for molecular graph message passing | CIN · ZINC-FULL | MAE: 0.036±0.002 -> 0.022±0.002 | Helped |
| Use induced-cycle cells with two message-passing layers | CIN · Mol-HIV | ROC-AUC: 79.05±1.32 -> 80.94±0.57 | Helped |
| Constrain the model to approximately 100k parameters (ablation) | CIN · ZINC, no edge features | MAE: 0.115±0.003 -> 0.139$\pm$0.008 | Hurt |
| Constrain the model to approximately 100k parameters (ablation) | CIN · ZINC, with edge features | MAE: 0.079±0.006 -> 0.094$\pm$0.004 | Hurt |
| Constrain the model to approximately 100k parameters (ablation) | CIN · ZINC-FULL | MAE: 0.022±0.002 -> 0.044$\pm$0.003 | Hurt |
| Constrain the model to approximately 100k parameters (ablation) | CIN · Mol-HIV | ROC-AUC: 80.94±0.57 -> 80.55$\pm$1.04 | Hurt |
| Use three cellular message-passing layers rather than depth proportional to ring size | CIN · RingTransfer | accuracy | Helped |
Closest papers by meaning
- Weisfeiler and Lehman Go Paths: Learning Topological Features via Path Complexes (AAAI 2024 · similarity 0.89)
- From Latent Graph to Latent Topology Inference: Differentiable Cell Complex Module (ICLR 2024 · similarity 0.87)
- Higher-Order Graph Convolutional Network with Flower-Petals Laplacians on Simplicial Complexes (AAAI 2024 · similarity 0.86)
- Topological Blindspots: Understanding and Extending Topological Deep Learning Through the Lens of Expressivity (ICLR 2025 · similarity 0.86)
- Beyond Message Passing: Neural Graph Pattern Machine (ICML 2025 · similarity 0.86)
- TopoTune: A Framework for Generalized Combinatorial Complex Neural Networks (ICML 2025 · 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.