Research map / Graph learning / GNN architectures
Robust Optimization As Data Augmentation for Large-Scale 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 |
|---|---|---|---|
| Add FLAG feature perturbations | GAT · ogbn-products | Test Acc: 79.45±0.59 -> 81.76±0.45 | Helped |
| Add FLAG feature perturbations | GraphSAGE · ogbl-ddi | Hits@20: 53.90 ±4.74 -> 63.31 ±6.06 | Helped |
| Add FLAG feature perturbations | GIN · ogbg-molhiv | Test ROC-AUC: 75.58±1.40 -> 76.54±1.14 | Helped |
| Use FLAG with virtual-node GCN | GCN-Virtual · ogbg-molhiv | Test ROC-AUC: 75.99±1.19 -> 75.45±1.58 | Hurt |
| Use FLAG with virtual-node GIN | GIN-Virtual · ogbg-molpcba | Test AP: 27.03±0.23 -> 28.34±0.38 | Helped |
| Replace FLAG with PGD adversarial augmentation (ablation) | GAT · ogbn-products | Test Acc: 80.96±0.41 vs 81.76±0.45 | Hurt |
| Replace FLAG with Free adversarial training (ablation) | GAT · ogbn-products | Test Acc: 79.42±0.84 vs 81.76±0.45 | Hurt |
| Reduce training epochs to match baseline runtime (ablation) | GAT + FLAG · ogbn-products | Test Acc: 81.76±0.45 -> 80.64±0.74 | Hurt |
| Use dual batch normalization with FLAG (ablation) | GCN + FLAG · ogbn-arxiv | Test Acc: 72.04±0.20 -> 72.11±0.23 | Helped |
| Add FLAG to Cluster mini-batch training (ablation) | GraphSAGE · ogbn-products | Test Acc: 78.97±0.33 -> 78.60±0.27 | Hurt |
| Add FLAG to SAINT mini-batch training (ablation) | GraphSAGE · ogbn-products | Test Acc: 79.08±0.24 -> 79.60±0.19 | Helped |
| Add FLAG on top of dropout (ablation) | GAT with dropout · ogbn-products | Test Acc: 79.45±0.59 -> 81.76±0.45 | Helped |
Closest papers by meaning
- Through the Dual-Prism: A Spectral Perspective on Graph Data Augmentation for Graph Classifications (AAAI 2025 · similarity 0.88)
- TRACI: A Data-centric Approach for Multi-Domain Generalization on Graphs (AAAI 2025 · similarity 0.88)
- Adversarial Contrastive Graph Augmentation with Counterfactual Regularization (AAAI 2025 · similarity 0.88)
- Graph Neural Network Generalization With Gaussian Mixture Model Based Augmentation (ICML 2025 · similarity 0.87)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.87)
- Edge Prompt Tuning for Graph Neural Networks (ICLR 2025 · similarity 0.87)
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.