Research map / Graph learning / GNN architectures
Data Augmentation for 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 |
|---|---|---|---|
| Use GAug-M learned deterministic edge modification | GCN · Cora | test micro-F1: 81.6$\pm$0.7 -> 83.5$\pm$0.4 | Helped |
| Use GAug-O learned stochastic edge augmentation | GCN · Cora | test micro-F1: 81.6$\pm$0.7 -> 83.6$\pm$0.5 | Helped |
| Use GAug-M learned deterministic edge modification | GCN · Flickr | test micro-F1: 61.2$\pm$0.4 -> 68.2$\pm$0.7 | Helped |
| Use GAug-O learned stochastic edge augmentation | GCN · PPI | test micro-F1: 43.4$\pm$0.2 -> 46.6$\pm$0.3 | Helped |
| Add stochastic graph sampling during training (ablation) | GAug-O on Cora · Cora | test micro-F1: 82.8$\pm$0.9 -> 83.6$\pm$0.5 (GCN); 81.2$\pm$0.8 -> 82.0$\pm$0.5 (GSAGE); 77.8$\pm$2.2 -> 82.2$\pm$0.8 (GAT); 76.9$\pm$1 | Helped |
| Use stochastic sampling instead of deterministic rounding (ablation) | GAug-O on Cora · Cora | test micro-F1: 82.5$\pm$0.5 -> 83.6$\pm$0.5 (GCN); 81.4$\pm$0.5 -> 82.0$\pm$0.5 (GSAGE); 81.3$\pm$1.1 -> 82.2$\pm$0.8 (GAT); 79.5$\pm$1 | Helped |
| Add the edge-prediction reconstruction loss (ablation) | GAug-O on Cora · Cora | test micro-F1: 82.8$\pm$0.8 -> 83.6$\pm$0.5 (GCN); 81.5$\pm$1.1 -> 82.0$\pm$0.5 (GSAGE); 81.9$\pm$0.8 -> 82.2$\pm$0.8 (GAT); 79.5$\pm$1 | Helped |
| Use GAug-M instead of GAug-M with inference on the original graph (ablation) | GCN · Flickr | test micro-F1: 61.6$\pm$0.6 -> 68.2$\pm$0.7 | Helped |
Closest papers by meaning
- Local Augmentation for Graph Neural Networks (ICML 2022 · similarity 0.93)
- Through the Dual-Prism: A Spectral Perspective on Graph Data Augmentation for Graph Classifications (AAAI 2025 · similarity 0.91)
- Graph Neural Network Generalization With Gaussian Mixture Model Based Augmentation (ICML 2025 · similarity 0.89)
- Edge Prompt Tuning for Graph Neural Networks (ICLR 2025 · similarity 0.88)
- Towards Bridging Generalization and Expressivity of Graph Neural Networks (ICLR 2025 · similarity 0.87)
- Integrating Co-Training with Edge Discrimination to Enhance Graph Neural Networks Under Heterophily (AAAI 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.