Research map / Graph learning / Graph self-supervised learning
Source-Free Graph Foundation Model Adaptation via Pseudo-Source Reconstruction
Liang Yang, Hui Ning, Jiaming Zhuo, Ziyi Ma, Chuan Wang, Wenning Wu, Zhen Wang
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 |
|---|---|---|---|
| use GCN backbone instead of GAT/SAGE/GIN (ablation) | SFGAR · Twitch EN→DE, FR→DE, FR→EN | accuracy: 61.74 -> 64.52 (GAT vs GCN, EN→DE) | Improved |
| add pseudo-source graph adversarial alignment (SFGAR) over GraphCTA | source-free graph DA · Twitch six tasks | accuracy: 63.05 -> 64.52 (EN→DE) | Improved |
| SFGAR vs GraphCTA on Elliptic | source-free graph DA · Elliptic S→M, S→E, M→E | accuracy: 85.52 -> 83.52 (S→M); 79.47 -> 80.18 (S→E) | Mixed |
| add both attribute loss and topology loss to pseudo-source generation (ablation) | SFGAR · Twitch/Elliptic | accuracy | Improved |
| increase number of GNN layers beyond 1 (ablation) | SFGAR · three datasets | accuracy | Hurt |
| vary pseudo-source nodes N' and edges E' (ablation) | SFGAR · hyperparameter study | accuracy: within 2% | No clear effect |
Closest papers by meaning
- Rethinking Propagation for Unsupervised Graph Domain Adaptation (AAAI 2024 · similarity 0.92)
- Revisiting, Benchmarking and Understanding Unsupervised Graph Domain Adaptation (NeurIPS 2024 · similarity 0.91)
- Smoothness Really Matters: A Simple Yet Effective Approach for Unsupervised Graph Domain Adaptation (AAAI 2025 · similarity 0.89)
- Domain Adaptive Unfolded Graph Neural Networks (AAAI 2025 · similarity 0.88)
- Revisiting Source-Free Domain Adaptation: Insights into Representativeness, Generalization, and Variety (CVPR 2025 · similarity 0.87)
- Towards Unsupervised Open-Set Graph Domain Adaptation via Dual Reprogramming (NeurIPS 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.
