Research map / Label-efficient and robust learning / Recognition and long-tail
Learning to Group Auxiliary Datasets for Molecule
Tinglin Huang, Ziniu Hu, Zhitao Ying
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 |
|---|---|---|---|
| Select auxiliary datasets using MolGroup rather than the strongest comparison baseline | GIN · 11 small-molecule target datasets | average relative improvement: 4.41% | Improved |
| Select auxiliary datasets using MolGroup rather than the strongest comparison baseline | Graphormer · 11 small-molecule target datasets | average relative improvement: 3.47% | Improved |
| Add PCBA as an auxiliary dataset (ablation) | GIN · BBBP | ROAUC: 66.62 -> 67.11 | Improved |
| Add PCBA as an auxiliary dataset (ablation) | GIN · ClinTox | ROAUC: 56.45 -> 57.77 | Improved |
| Add PCBA as an auxiliary dataset (ablation) | GIN · Tox21 | ROAUC: 74.23 -> 74.81 | Improved |
| Add PCBA as an auxiliary dataset (ablation) | GIN · BACE | ROAUC | Hurt |
| Remove target-dataset meta-gradient guidance from routing optimization (ablation) | MolGroup · BBBP | affinity discrimination | Hurt |
| Use a lower task-structure balance coefficient than 0.9 (ablation) | MolGroup · BBBP, Tox21, and Lipo | auxiliary affinity scores | Hurt |
| Increase filtering rounds from 1 to 3 (ablation) | MolGroup · BBBP | ROAUC: 67.94_{0.018} -> 68.36_{0.016} | Improved |
| Train with all auxiliary datasets using the final soft routing gate instead of filtering negative datasets (ablation) | MolGroup · BBBP | ROAUC: 68.36_{0.016} -> 66.83_{0.030} | Hurt |
| Add more top-affinity auxiliary datasets (ablation) | MolGroup · FreeSolv | RMSE: 3.842_{1.579} -> 3.116_{0.279} | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Task2vec — Task2Vec: Task Embedding for Meta-Learning (ICCV 2019) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- S-MolSearch: 3D Semi-supervised Contrastive Learning for Bioactive Molecule Search (NeurIPS 2024 · similarity 0.86)
- UniMatch: Universal Matching from Atom to Task for Few-Shot Drug Discovery (ICLR 2025 · similarity 0.86)
- Empower Structure-Based Molecule Optimization with Gradient Guided Bayesian Flow Networks (ICML 2025 · similarity 0.85)
- Rethinking the generalization of drug target affinity prediction algorithms via similarity aware evaluation (ICLR 2025 · similarity 0.84)
- A Systematic Evaluation of Co-folding Model Representations for Small-Molecule Learning (NeurIPS 2026 · similarity 0.84)
- Offline Model-based Optimization for Real-World Molecular Discovery (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.
