Research map / Graph learning / Graph self-supervised learning
Revisiting Semi-Supervised Learning with Graph Embeddings
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 |
|---|---|---|---|
| Jointly train graph-context prediction with label prediction instead of learning graph embeddings separately (ablation) | Planetoid-G vs GraphEmb · Citeseer, Cora, Pubmed | accuracy | Helped |
| Jointly train graph-context prediction with label prediction instead of learning graph embeddings separately (ablation) | Planetoid-G vs GraphEmb · NELL | accuracy: +5.0%, +3.2%, +3.8% | Helped |
| Add feature-conditioned embeddings (Planetoid-I rather than Planetoid-G) (ablation) | Planetoid · NELL | accuracy | Hurt |
| Use the inductive formulation rather than the transductive formulation (ablation) | Planetoid · Citeseer and Pubmed | accuracy | Helped |
| Use the transductive formulation rather than the inductive formulation (ablation) | Planetoid · Cora | accuracy: 14.5% | Helped |
| Use the inductive formulation rather than the transductive formulation (ablation) | Planetoid · DIEL | Recall@k | No clear effect |
| Add feature vectors to Planetoid-T (ablation) | Planetoid-T · NELL | accuracy | No clear effect |
| Replace graph Laplacian regularization with graph-context prediction | Planetoid-I vs ManiReg and SemiEmb · Text classification | accuracy: up to 6.1% | Helped |
| Use Planetoid-I rather than the best other inductive method | Planetoid-I · NELL labeling rates 0.1, 0.01, and 0.001 | accuracy: 4.8%, 16.0%, and 18.7% | Helped |
Closest papers by meaning
- Semi-Supervised Classification with Graph Convolutional Networks (ICLR 2017 · similarity 0.91)
- A Flexible Generative Framework for Graph-based Semi-supervised Learning (NeurIPS 2019 · similarity 0.90)
- Data driven semi-supervised learning (NeurIPS 2021 · similarity 0.89)
- Label Efficient Semi-Supervised Learning via Graph Filtering (CVPR 2019 · similarity 0.89)
- Learning Graph Representations with Embedding Propagation (NeurIPS 2017 · similarity 0.89)
- Hypergraph-enhanced Dual Semi-supervised Graph Classification (ICML 2024 · similarity 0.88)
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.