Research map / Graph learning / GNN architectures
Convergence of Invariant Graph 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 edge-weight inputs and compare graph outputs with graphon outputs | k-IGN · Edge weight continuous model, fixed sampling | convergence | Helped |
| Use edge-weight inputs with random graph sampling | k-IGN · Edge weight continuous model, random sampling | convergence | Helped |
| Feed the sampled 0-1 adjacency matrix directly, without edge probability estimation | general IGN · Edge probability discrete model | convergence | Hurt |
| Estimate edge probabilities before applying an IGN in the IGN{} subset | IGN{} · Edge probability discrete model | convergence | Helped |
| Use the partition-norm to assess linear-layer stability | k-IGN linear equivariant layers · Graph and graphon inputs | stability | Helped |
| Use IGN{} rather than unrestricted IGN | IGN · Compact domain, spectral GNN approximation | approximation capability | Helped |
| Increase the sampled graph size | Untrained 2-IGN with random weights · Deterministic and random graphon sampling | output difference | Helped |
| Use a 0-1 adjacency matrix instead of edge weights | Untrained 2-IGN with random weights · Graphon sampling experiments | output difference | Hurt |
| Apply edge probability estimation before feeding the graph to IGN | Untrained 2-IGN with random weights · Graphon sampling experiments | output difference | Helped |
Closest papers by meaning
- Higher-Order Graphon Neural Networks: Approximation and Cut Distance (ICLR 2025 · similarity 0.91)
- Almost Surely Asymptotically Constant Graph Neural Networks (NeurIPS 2024 · similarity 0.88)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 2025 · similarity 0.87)
- Position: Spectral GNNs Rely Less on Graph Fourier Basis than Conceived (ICML 2025 · similarity 0.87)
- Homomorphism Expressivity of Spectral Invariant Graph Neural Networks (ICLR 2025 · similarity 0.87)
- MuseGNN: Forming Scalable, Convergent GNN Layers that Minimize a Sampling-Based Energy (ICLR 2025 · similarity 0.87)
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.