Research map / Graph learning / Graph self-supervised learning
Subgraph Federated Learning with Missing Neighbor Generation
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 |
|---|---|---|---|
| add federation (FedAvg across owners) to GraphSage instead of local training | GraphSage · Cora M=10 | accuracy: 0.2798 -> 0.8626 | Helped |
| add NeighGen missing-neighbor generation to FedSage | FedSage · Citeseer M=3 | accuracy: 0.7241 -> 0.7454 | Helped |
| add NeighGen to FedSage | FedSage · Cora M=3 | accuracy: 0.8656 -> 0.8686 | Helped |
| add NeighGen to FedSage | FedSage · MSAcademic M=3 | accuracy: 0.9327 -> 0.9359 | Helped |
| add NeighGen to local GraphSage training (LocSage+ vs LocSage) | LocSage · Cora M=3 | accuracy: 0.5762 -> 0.5644 | No clear effect |
| use locally specialized NeighGens instead of a globally shared FedAvg NeighGen (ablation) | FedSage+ · MSAcademic M=3 | accuracy: 0.9210 -> 0.9414 | Helped |
| use locally specialized NeighGens instead of a globally shared FedAvg NeighGen (ablation) | FedSage+ · Cora M=3 | accuracy: 0.8619 -> 0.8686 | Helped |
| add a globally shared FedAvg NeighGen to FedSage (ablation) | FedSage · Cora/CiteSeer/MSAcademic M=3 | accuracy: 0.8656 -> 0.8619 (Cora) | Hurt |
| vary hiding ratio h to extremes (1% or 30%) (ablation) | FedSage+ · hyperparameter study | accuracy | Hurt |
| set cross-subgraph loss weight alpha to extremes (0.1 or 10) instead of 1 (ablation) | FedSage+ · hyperparameter study | accuracy | Hurt |
Closest papers by meaning
- Subgraph Federated Learning for Local Generalization (ICLR 2025 · similarity 0.90)
- Decoupled Subgraph Federated Learning (ICLR 2025 · similarity 0.89)
- Modeling Inter-Intra Heterogeneity for Graph Federated Learning (AAAI 2025 · similarity 0.88)
- Federated Node-Level Clustering Network with Cross-Subgraph Link Mending (ICML 2025 · similarity 0.87)
- Personalized Subgraph Federated Learning with Differentiable Auxiliary Projections (NeurIPS 2025 · similarity 0.87)
- Federated Graph-Level Clustering Network (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.