Research map / Graph learning / GNN architectures
Elastic Graph Neural 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 |
|---|---|---|---|
| Add ℓ1-based smoothing to ℓ2 smoothing | Elastic GNN · Cora, CiteSeer, PubMed, Coauthor Physics, Amazon Photo | accuracy (%): 79.6 -> 82.7; 68.9 -> 70.9; 77.6 -> 79.4; 93.3 -> 94.2; 90.3 -> 91.3 | Helped |
| Add ℓ1-based smoothing to ℓ2 smoothing | Elastic GNN · Coauthor CS, Amazon Computers | accuracy (%): 91.6 -> 92.5; 79.8 -> 80.7 | No clear effect |
| Add ℓ1-based smoothing (ablation) | ℓ2-only Elastic GNN · Cora, 15% perturbation | accuracy (%): 72.2 -> 77.2 | Helped |
| Add ℓ2-based smoothing (ablation) | ℓ21-only Elastic GNN · Cora, 15% perturbation | accuracy (%): 75.7 -> 77.2 | Helped |
| Use ℓ21 coupling instead of independent ℓ1 penalties (ablation) | Elastic GNN with ℓ1+ℓ2 · Cora, 15% perturbation | accuracy (%): 76.9 -> 77.2 | Helped |
| Use ℓ21 coupling instead of independent ℓ1 penalties (ablation) | Elastic GNN · adversarial graph attacks | accuracy (%) | Helped |
| Combine ℓ21 and ℓ2 smoothing rather than use either alone (ablation) | Elastic GNN · adversarial graph attacks | accuracy (%) | Helped |
| Use ℓ21 smoothing rather than ℓ2-only smoothing (ablation) | Elastic GNN · adversarial graph attacks | accuracy (%) | Helped |
| Use ℓ21+ℓ2 smoothing rather than ℓ2 smoothing (ablation) | Elastic GNN · Cora | wrong-edge/correct-edge difference ratio: 1.57 -> 2.03 | Helped |
| Use ℓ21+ℓ2 smoothing rather than ℓ2 smoothing (ablation) | Elastic GNN · Cora, CiteSeer, PubMed | sparsity ratio: 2% -> 37%; 16% -> 74%; 11% -> 42% | Helped |
| Increase propagation steps K (ablation) | EMP · Cora, CiteSeer, PubMed | classification accuracy | Helped |
| Run EMP message passing iterations (ablation) | Elastic GNN · elastic estimator objective | objective value | Helped |
Closest papers by meaning
- Measuring and Relieving the Over-Smoothing Problem for Graph Neural Networks from the Topological View (AAAI 2020 · similarity 0.88)
- Moderate Message Passing Improves Calibration: A Universal Way to Mitigate Confidence Bias in Graph Neural Networks (AAAI 2024 · similarity 0.85)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.85)
- What do Graph Neural Networks learn? Insights from Tropical Geometry (NeurIPS 2024 · similarity 0.85)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.85)
- Almost Surely Asymptotically Constant Graph Neural Networks (NeurIPS 2024 · similarity 0.85)
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.