Research map / Graph learning / Graph self-supervised learning
Property-Aware Relation Networks for Few-Shot Molecular Property Prediction
Yaqing Wang, ABULIKEMU ABUDUWEILI, quanming yao, Dejing Dou
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 |
|---|---|---|---|
| add property-aware embedding, adaptive relation graph and neighbor-alignment regularizer (PAR) over scratch-trained baselines | GIN encoder · Tox21 10-shot | ROC-AUC: 81.21 -> 82.06 | Improved |
| add PAR modules on top of pretrained Pre-GNN (Pre-PAR) | Pre-GNN · Tox21 10-shot | ROC-AUC: 82.14 -> 84.93 | Improved |
| add PAR modules on top of pretrained Pre-GNN (Pre-PAR) | Pre-GNN · SIDER 10-shot | ROC-AUC: 73.96 -> 78.08 | Improved |
| add PAR modules on top of pretrained Pre-GNN (Pre-PAR) | Pre-GNN · ToxCast 10-shot | ROC-AUC: 73.68 -> 75.12 | Improved |
| use PAR instead of IterRefLSTM | GIN encoder · Tox21 1-shot | ROC-AUC: 80.97 -> 80.46 | No clear effect |
| use Pre-PAR instead of Meta-MGNN | pretrained GIN · MUV 10-shot | ROC-AUC: 68.99 -> 69.96 | Improved |
| add property-aware embedding function (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
| add adaptive relation graph learning (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
| learn similarity function instead of fixed cosine similarity (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
| reduce relation graph to KNN graph (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
| add neighbor alignment regularizer (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
| fine-tune only property-aware parameters within tasks instead of all parameters (ablation) | PAR / Pre-PAR · Tox21 10-shot | ROC-AUC | Improved |
Closest papers by meaning
- ReCoG: Relational and Compact Context Graph Learning for Few-shot Molecular Property Prediction (ICML 2026 · similarity 0.91)
- Association Pattern-enhanced Molecular Representation Learning (AAAI 2025 · similarity 0.88)
- Knowledge Enhanced Representation Learning for Drug Discovery (AAAI 2024 · similarity 0.87)
- MIPT: Multilevel Informed Prompt Tuning for Robust Molecular Property Prediction (ICML 2025 · similarity 0.87)
- Pin-Tuning: Parameter-Efficient In-Context Tuning for Few-Shot Molecular Property Prediction (NeurIPS 2024 · similarity 0.86)
- Foundation Molecular Grammar: Multi-Modal Foundation Models Induce Interpretable Molecular Graph Languages (ICML 2025 · similarity 0.86)
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.
