Research map / Graph learning / GNN architectures
Graph Neural Networks Exponentially Lose Expressive Power for Node Classification
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 |
|---|---|---|---|
| Scale weights so their maximum singular value is above the information-loss threshold | GCN · Noisy Cora | test accuracy | Helped |
| Set maximum singular value to 10 instead of leaving weights unnormalized | GCN · Noisy Cora | test accuracy | Helped |
| Increase the relative magnitude of the output component perpendicular to the invariant space | GCN · Noisy Cora | correlation with prediction accuracy: R=0.545 | Helped |
| Increase the relative magnitude of the output component perpendicular to the invariant space | GCN · Noisy Cora, excluding one-layer GCNs | correlation with prediction accuracy: 0.827 | Helped |
| Use a non-negative eigenvector basis for the invariant space rather than a sign-mixed eigenvector (ablation) | One-step transition · Synthetic two-node case | distance to invariant space | Helped |
| Use a sign-mixed eigenvector basis for the invariant space rather than a non-negative basis (ablation) | One-step transition · Synthetic two-node case | distance to invariant space | Hurt |
| Increase the product of maximum singular value and spectral contraction factor (ablation) | GCN on synthetic Erdős–Rényi graphs · Synthetic Erdős–Rényi graphs | distance to invariant space | Mixed |
Closest papers by meaning
- Understanding the Representation Power of Graph Neural Networks in Learning Graph Topology (NeurIPS 2019 · similarity 0.92)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 2025 · similarity 0.91)
- Almost Surely Asymptotically Constant Graph Neural Networks (NeurIPS 2024 · similarity 0.90)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.90)
- On the Interplay between Graph Structure and Learning Algorithms in Graph Neural Networks (ICML 2025 · similarity 0.90)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.90)
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.