Research map / Graph learning / Graph self-supervised learning
Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs
Check 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 |
|---|---|---|---|
| Condition filters on edge labels rather than using uniform labels (ablation) | ECC · NCI1 | accuracy: 76.82 -> 83.80 | Helped |
| Condition filters on edge labels rather than using uniform labels (ablation) | ECC · MUTAG | accuracy: 76.11 -> 89.44 | Helped |
| Increase convolution radii to 1.5ρ (ablation) | ECC · Sydney Urban Objects | weighted mean F1: 78.4 -> 76.9 | Hurt |
| Increase convolution radii to 2ρ (ablation) | ECC · Sydney Urban Objects | weighted mean F1: 78.4 -> 74.4 | Hurt |
| Add identity skip-connection to ECC 1.5ρ (ablation) | ECC 1.5ρ · Sydney Urban Objects | weighted mean F1: 76.9 -> 79.5 | Helped |
| Average predictions over 12 orientations (ablation) | ECC · ModelNet10 | mean class accuracy: 89.3 -> 90.0 | Helped |
| Average predictions over 12 orientations (ablation) | ECC · ModelNet40 | mean class accuracy: 82.4 -> 83.2 | Helped |
| Discard black points from input graphs (ablation) | ECC · MNIST | test accuracy: 99.14 -> 99.14 | No clear effect |
| Use one-hot offset labels and single-layer filter networks (ablation) | ECC · MNIST | test accuracy: 99.14 -> 99.37 | Helped |
| Add identity skip-connections (ablation) | ECC · graph classification, NCI1 | accuracy: 83.80 -> 83.24 | Hurt |
| Add vertex degrees to edge labels (ablation) | ECC · ENZYMES | accuracy: 50.00 -> 55.00 | Helped |
| Add vertex degrees to edge labels (ablation) | ECC · NCI109 | accuracy: 81.87 -> 83.03 | Mixed |
Closest papers by meaning
- Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering (NeurIPS 2016 · similarity 0.91)
- Simplifying Graph Convolutional Networks (ICML 2019 · similarity 0.91)
- Learning Convolutional Neural Networks for Graphs (ICML 2016 · similarity 0.90)
- Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks (NeurIPS 2019 · similarity 0.89)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.89)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 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.