Research map / Graph learning / Graph self-supervised learning
Fused Gromov-Wasserstein Graph Mixup for Graph-level Classifications
Xinyu Ma, Xu Chu, Yasha Wang, Yang Lin, Junfeng Zhao, Liantao Ma, Wenwu Zhu
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 FGWMixup to vGCN training | vGCN · PROTEINS | accuracy (%): 74.75(2.60) -> 76.01(3.19) | Improved |
| Add FGWMixup to vGCN training | vGCN · NCI1 | accuracy (%): 76.91(1.80) -> 78.37(2.40) | Improved |
| Use FGWMixup instead of G-Mixup | vGCN · IMDB-M | accuracy (%): 49.60(3.90) -> 50.80(4.06) | Improved |
| Add FGWMixup to Graphormer | Graphormer · NCI1 | accuracy (%): 61.56(3.70) -> 66.45(2.58) | Improved |
| Add FGWMixup to GraphormerGD | GraphormerGD · IMDB-M | accuracy (%): 47.47(2.98) -> 48.87(4.03) | Improved |
| Use single-loop relaxed FGW solver (FGWMixup_*) instead of strict CG solver | FGWMixup · mixup time | speedup: 2.03x to 3.46x | Improved |
| Use FGWMixup_* instead of FGWMixup | vGCN · NCI1 | accuracy (%): 78.37(2.40) -> 78.47(1.74) | No clear effect |
| Use FGWMixup_* instead of FGWMixup | Graphormer · NCI1 | accuracy (%): 66.45(2.58) -> 64.26(3.25) | Hurt |
| Set alpha to 1.0 (GW only, drop node features) instead of 0.95 (ablation) | vGIN-FGWMixup · NCI1 | accuracy (%): 78.32(2.65) -> 75.91(2.93) | Hurt |
| Set alpha to 1.0 instead of 0.95 (ablation) | vGCN-FGWMixup · PROTEINS | accuracy (%): 76.01(3.19) -> 74.40(3.57) | Hurt |
| Use fixed mixup graph size instead of adaptive size (ablation) | FGWMixup · NCI1, PROTEINS | accuracy (%): overall performance decay of over 1% | Hurt |
| Add FGWMixup under 60% label corruption | vGCN · IMDB-B | accuracy (%): 47.90(4.30) -> 50.80(3.19) | Improved |
Closest papers by meaning
- Graph Mixup on Approximate Gromov–Wasserstein Geodesics (ICML 2024 · similarity 0.89)
- AGMixup: Adaptive Graph Mixup for Semi-supervised Node Classification (AAAI 2025 · similarity 0.88)
- IntraMix: Intra-Class Mixup Generation for Accurate Labels and Neighbors (NeurIPS 2024 · similarity 0.87)
- Fine-Tuning Graph Neural Networks by Preserving Graph Generative Patterns (AAAI 2024 · similarity 0.86)
- Graph Invariant Learning with Subgraph Co-mixup for Out-of-Distribution Generalization (AAAI 2024 · similarity 0.86)
- Graph Distillation with Eigenbasis Matching (ICML 2024 · similarity 0.86)
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.
