Research map / Graph learning / GNN architectures
Reliable Graph Neural Networks via Robust Aggregation
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 |
|---|---|---|---|
| replace mean aggregation with Soft Medoid (T=0.2) on GDC | GDC · Cora ML | accumulated certifications: 1.98 -> 5.60 | Helped |
| replace mean aggregation with Soft Medoid (T=0.2) on GDC | GDC · Citeseer | accumulated certifications: 1.24 -> 4.69 | Helped |
| use Soft Medoid instead of hard Medoid (ablation) | GDC · Cora ML | accumulated certifications: 4.05 -> 5.60 | Helped |
| use Soft Medoid instead of hard Medoid (ablation) | GDC · Cora ML | clean accuracy: 0.724 -> 0.823 | Helped |
| use Soft Medoid instead of dimension-wise median (ablation) | GDC · Cora ML | accumulated certifications: 2.38 -> 5.60 | Helped |
| add GDC preprocessing to Soft Medoid GCN (ablation) | Soft Medoid · Cora ML | accumulated certifications: 1.86 -> 4.31 | Helped |
| add GDC alone to GCN (no Soft Medoid) (ablation) | GCN · Cora ML | accumulated certifications: 1.84 -> 1.98 | No clear effect |
| lower temperature T from 1.0 to 0.2 (ablation) | SM GDC · Cora ML | accumulated certifications: 4.31 -> 5.60; accuracy 0.823 -> 0.770 | Mixed |
| use Soft Medoid GDC vs defenses (SVD, Jaccard, RGCN) | SM GDC · Cora ML | accumulated certifications: 0.86 -> 5.60 | Helped |
Closest papers by meaning
- Robust Graph Neural Networks via Unbiased Aggregation (NeurIPS 2024 · similarity 0.93)
- Robustness of Graph Neural Networks at Scale (NeurIPS 2021 · similarity 0.91)
- Bounding the Expected Robustness of Graph Neural Networks Subject to Node Feature Attacks (ICLR 2024 · similarity 0.89)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.89)
- Robust Graph Condensation via Classification Complexity Mitigation (NeurIPS 2025 · similarity 0.88)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.88)
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.