Research map / Graph learning / GNN architectures
On Over-Squashing in Message Passing Neural Networks: The Impact of Width, Depth, and Topology
Francesco Di Giovanni, Lorenzo Giusti, Federico Barbero, Giulia Luise, Pietro Lio, Michael M. Bronstein
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| increase hidden width | MPNNs · Graph transfer tasks: CrossedRing, Ring, CliquePath | task performance | Improved |
| increase hidden width and weight magnitude | MPNNs · Long-range information propagation | generalization | Hurt |
| increase depth to propagate information between distant nodes | MPNNs · Long-range dependencies | information sensitivity | No clear effect |
| use greater depth in the deep regime | MPNNs · Deep MPNNs | loss gradient magnitude | Hurt |
| use a topology with lower commute time between task-relevant nodes | MPNNs · Graph transfer tasks | task difficulty | Improved |
| rewire the graph to increase its Cheeger constant | MPNNs · Graph rewiring | Jacobian obstruction | Improved |
| add edges that reduce effective resistance between nodes | MPNNs · Spatial graph rewiring | information exchange | Improved |
Closest papers by meaning
- PANDA: Expanded Width-Aware Message Passing Beyond Rewiring (ICML 2024 · similarity 0.91)
- Mitigating Over-squashing without Rewiring: A Sheaf Effective Resistance Perspective (NeurIPS 2026 · similarity 0.91)
- Effects of Dropout on Performance in Long-range Graph Learning Tasks (NeurIPS 2025 · similarity 0.90)
- Over-squashing in Spatiotemporal Graph Neural Networks (NeurIPS 2025 · similarity 0.89)
- Mitigating Over-Squashing in Graph Neural Networks by Spectrum-Preserving Sparsification (ICML 2025 · similarity 0.89)
- Deeper with Riemannian Geometry: Overcoming Oversmoothing and Oversquashing for Graph Foundation Models (NeurIPS 2025 · similarity 0.89)
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.
