Research map / Graph learning / Knowledge and temporal graphs
Few-shot Relation Extraction via Bayesian Meta-learning on Relation Graphs
Meng Qu, Tianyu Gao, Louis-Pascal Xhonneux, Jian Tang
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 REGRAB instead of the strongest listed baseline | Few-shot relation extraction with BERT_BASE · FewRel test, 5-Way 1-Shot | accuracy: 89.80 -> 90.30 | Improved |
| use REGRAB instead of the strongest listed baseline | Few-shot relation extraction with BERT_BASE · FewRel test, 10-Way 5-Shot | accuracy: 88.64 -> 89.93 | Improved |
| remove the graph-based prior (comparison: add the graph-based prior) (ablation) | REGRAB · FewRel validation, 5-Way 1-Shot | accuracy: 85.82 -> 87.95 | Improved |
| remove the graph-based prior (comparison: add the graph-based prior) (ablation) | REGRAB · FewRel validation, 10-Way 1-Shot | accuracy: 77.70 -> 80.26 | Improved |
| replace amortized variational inference with Langevin dynamics (ablation) | REGRAB · FewRel validation, 5-Way 1-Shot | accuracy: 85.74 -> 87.95 | Improved |
| replace amortized variational inference with Langevin dynamics (ablation) | REGRAB · FewRel validation, 10-Way 1-Shot | accuracy: 77.41 -> 80.26 | Improved |
| use Euclidean distance instead of dot product (ablation) | REGRAB · FewRel test, 5-Way 1-Shot and 10-Way 1-Shot | accuracy: 90.30 -> 86.74; 84.09 -> 78.56 | Hurt |
| use Euclidean distance instead of dot product (ablation) | REGRAB · FewRel test, 5-Way 5-Shot and 10-Way 5-Shot | accuracy: 94.25 -> 94.34; 89.93 -> 88.95 | Mixed |
Closest papers by meaning
- Few-Shot, No Problem: Descriptive Continual Relation Extraction (AAAI 2025 · similarity 0.88)
- Synergistic Anchored Contrastive Pre-training for Few-Shot Relation Extraction (AAAI 2024 · similarity 0.87)
- MoEMeta: Mixture-of-Experts Meta Learning for Few-Shot Relational Learning (NeurIPS 2025 · similarity 0.86)
- A Hierarchical Bayesian Model for Few-Shot Meta Learning (ICLR 2024 · similarity 0.86)
- Enhancing Low-Resource Relation Representations through Multi-View Decoupling (AAAI 2024 · similarity 0.85)
- DocKS-RAG: Optimizing Document-Level Relation Extraction through LLM-Enhanced Hybrid Prompt Tuning (ICML 2025 · similarity 0.84)
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.
