Research map / Graph learning / Graph self-supervised learning
Node Classification on Graphs with Few-Shot Novel Labels via Meta Transformed Network Embedding
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 MetaTNE rather than Planetoid | few-shot novel-label node classification · BlogCatalog, 10 positive/20 negative support | F1: 0.4657 -> 0.5380 | Helped |
| Use MetaTNE rather than Planetoid | few-shot novel-label node classification · Flickr, 10 positive/20 negative support | F1: 0.6638 -> 0.7118 | Helped |
| Use MetaTNE rather than Planetoid | few-shot novel-label node classification · PPI, 10 positive/20 negative support | F1: 0.4672 -> 0.5188 | Helped |
| Use MetaTNE rather than Planetoid | few-shot novel-label node classification · Mashup, 10 positive/20 negative support | F1: 0.4825 -> 0.5764 | Helped |
| Use MetaTNE rather than Planetoid | few-shot novel-label node classification · Flickr, 10 positive/40 negative support | AUC: 0.8728 -> 0.8505 | Hurt |
| Add the embedding transformation function (ablation) | MetaTNE · BlogCatalog, 10 positive/20 negative support | F1: 0.5028 -> 0.5380 | Helped |
| Add the embedding transformation function (ablation) | MetaTNE · PPI, 10 positive/20 negative support | F1: 0.4851 -> 0.5188 | Helped |
| Use query-specific, separately transformed positive and negative support sets instead of joint self-attention over all nodes (ablation) | MetaTNE · BlogCatalog, 10 positive/20 negative support | F1: 0.5020 -> 0.5380 | Helped |
| Use query-specific, separately transformed positive and negative support sets instead of joint self-attention over all nodes (ablation) | MetaTNE · PPI, 10 positive/20 negative support | F1: 0.5011 -> 0.5188 | Helped |
| Schedule optimization between structural and meta-learning modules rather than optimizing a weighted total loss (ablation) | MetaTNE · BlogCatalog, 10 positive/20 negative support | F1: 0.5205 -> 0.5380 | Helped |
| Schedule optimization between structural and meta-learning modules rather than optimizing a weighted total loss (ablation) | MetaTNE · PPI, 10 positive/20 negative support | F1: 0.4980 -> 0.5188 | Helped |
| Jointly train structural embeddings and meta-learning rather than fixing embeddings after initial structural training (ablation) | MetaTNE · BlogCatalog, 10 positive/20 negative support | F1: 0.4748 -> 0.5380 | Helped |
Closest papers by meaning
- Meta Propagation Networks for Graph Few-shot Semi-supervised Learning (AAAI 2022 · similarity 0.91)
- Graph Few-shot Learning with Task-specific Structures (NeurIPS 2022 · similarity 0.91)
- Graph Few-Shot Learning via Knowledge Transfer (AAAI 2020 · similarity 0.90)
- NLGT: Neighborhood-based and Label-enhanced Graph Transformer Framework for Node Classification (AAAI 2025 · similarity 0.89)
- Node Identifiers: Compact, Discrete Representations for Efficient Graph Learning (ICLR 2025 · similarity 0.88)
- Leveraging Large Language Models for Node Generation in Few-Shot Learning on Text-Attributed Graphs (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.