Research map / Graph learning / Graph self-supervised learning
Is Graph Mixup Beneficial? Investigating Interpolation And Empirical Performance of Graph Mixup Methods
Simon Forbat, Rainer Gemulla
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 graph mixup (any evaluated method) | GCN/GIN graph classifiers · TUDataset (6 datasets) | accuracy (%) | No clear effect |
| add GED-Mixup (optimal alignment) to training | GCN/GIN · TUDataset pooled (A3 fixed data) | accuracy (%): 70.29 -> 70.54 | No clear effect |
| add SubMix | GCN/GIN · TUDataset pooled (A1) | accuracy (%): 70.29 -> 70.69 | No clear effect |
| add S-Mixup | GCN/GIN · TUDataset pooled (A3 fixed data) | accuracy (%): 70.29 -> 68.11 | Hurt |
| add GeoMix | GCN/GIN · TUDataset pooled (A3 fixed data) | accuracy (%): 70.29 -> 69.11 | Hurt |
| add Embedding Mixup | GCN/GIN · TUDataset pooled (A1) | accuracy (%): 70.29 -> 69.49 | No clear effect |
| use optimal GED alignment instead of default adjacency ordering (If-Mixup) | mixup interpolation · TUDataset | mIE: 0.91 -> 0.02 | Improved |
| inject interpolation error into GED-Mixup (mIE 0.000 to 1.000) (ablation) | GED-Mixup with GCN · IMDB-M | accuracy (%): 46.16 -> 35.51 | Hurt |
| use methods with low interpolation error (GED-Mixup, SubMix, If-Mixup) instead of high-error methods | graph mixup · TUDataset pooled | accuracy (%): p = 5.57 × 10−02, 7.61 × 10−06, 1.40 × 10−13 | Improved |
Closest papers by meaning
- AGMixup: Adaptive Graph Mixup for Semi-supervised Node Classification (AAAI 2025 · similarity 0.87)
- Tailoring Mixup to Data for Calibration (ICLR 2025 · similarity 0.86)
- Selective Mixup Helps with Distribution Shifts, But Not (Only) because of Mixup (ICML 2024 · similarity 0.85)
- IntraMix: Intra-Class Mixup Generation for Accurate Labels and Neighbors (NeurIPS 2024 · similarity 0.85)
- Graph Invariant Learning with Subgraph Co-mixup for Out-of-Distribution Generalization (AAAI 2024 · similarity 0.85)
- Pushing Boundaries: Mixup's Influence on Neural Collapse (ICLR 2024 · similarity 0.84)
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.
