Research map / Graph learning / Graph self-supervised learning
Transfer Learning for Deep Learning on Graph-Structured Data
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 SGE instead of CoGE for graph generation (ablation) | SCNN GC8-FC500 Lbasic · AG | classification accuracy: 0.89 -> 0.91 | Helped |
| Use SGE instead of CoGE for graph generation (ablation) | SCNN GC8-FC500 Lbasic · DBP | classification accuracy: 0.95 -> 0.97 | Helped |
| Use SGE instead of CoGE for graph generation (ablation) | SCNN GC8-GC8-FC1K Lbasic · YELP | classification accuracy: 0.92 -> 0.92 | No clear effect |
| Change Laplacian type among Lbasic, Lrw, Lrwr (ablation) | SCNN · AG/DBP/YELP/AMAZ | classification accuracy: 0.91 -> 0.89 (AG, SGE Lbasic vs Lrwr) | No clear effect |
| Add a second graph convolution layer (GC8-GC8-FC1K vs GC8-FC1K) (ablation) | SCNN SGE Lbasic · AMAZ | classification accuracy: 0.88 -> 0.88 | No clear effect |
| Transfer conv/pooling layers from source and fine-tune only FC layer, high graph similarity | SCNN · Synthetic high-similarity pair | test accuracy | Helped |
| Transfer to a target domain with low graph similarity | SCNN · Synthetic low-similarity pair | test accuracy | Hurt |
| Transfer between YELP and AMAZ (highest similarity) | SCNN · YELP/AMAZ | test accuracy: comparable (YELP), 5-8% lower (AMAZ) | Mixed |
| Reuse source-trained model with transfer instead of training from scratch | SCNN · text classification corpora | training time: more than 10% reduction in average training time | Helped |
Closest papers by meaning
- Contextual Structure Knowledge Transfer for Graph Neural Networks (AAAI 2025 · similarity 0.89)
- One for All: Universal Topological Primitive Transfer for Graph Structure Learning (NeurIPS 2025 · similarity 0.88)
- Network Structure and Transfer Behaviors Embedding via Deep Prediction Model (AAAI 2019 · similarity 0.88)
- Diffusion-Convolutional Neural Networks (NeurIPS 2016 · similarity 0.88)
- How Much Can Transfer? BRIDGE: Bounded Multi-Domain Graph Foundation Model with Generalization Guarantees (ICML 2025 · similarity 0.86)
- GraphBridge: Towards Arbitrary Transfer Learning in GNNs (ICLR 2025 · 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.