Research map / Probabilistic and scientific modeling / Molecules and proteins
Understanding the Limitations of Deep Models for Molecular property prediction: Insights and Solutions
Jun Xia, Lecheng Zhang, Xiao Zhu, Yue Liu, Zhangyang Gao, Bozhen Hu, Cheng Tan, Jiangbin Zheng, Siyuan Li, Stan Z. Li
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 Independent Feature Mapping (IFM) | MLP · FreeSolv | RMSE: 1.046 -> 0.862 | Improved |
| add IFM | MLP · BACE | AUC-ROC: 0.887 -> 0.915 | Improved |
| add IFM | GCN · HIV | AUC-ROC: 0.834 -> 0.862 | Improved |
| add IFM | MLP · MUV | AUC-PRC: 0.018 -> 0.052 | Improved |
| add IFM | MLP · ClinTox | AUC-ROC: 0.890 -> 0.941 | Improved |
| add IFM | GNN models · activity cliff datasets | RMSEc: 0.807 -> 0.715 | Improved |
| replace sinusoidal mapping with IFM (ablation) | MLP · FreeSolv | RMSE: 1.033 -> 0.862 | Improved |
| replace Gaussian mapping with IFM (ablation) | GCN · ESOL | RMSE: 0.766 -> 0.728 | Improved |
| replace Gaussian mapping with IFM (ablation) | MLP · BBBP | AUC-ROC: 0.923 -> 0.937 | Improved |
| apply IFM to pre-trained encoder | AttrMasking GCN · Tox21 | AUC-ROC: 0.745 -> 0.758 | Improved |
| apply IFM to pre-trained encoder | Mole-BERT GIN · Sider | AUC-ROC: 0.628 -> 0.627 | No clear effect |
| smooth labels with k-nearest neighbors (10-smooth, 20-smooth) (ablation) | deep models · ESOL, Lipop, QM7 | RMSE/MAE | Improved |
Closest papers by meaning
- BOOM: Benchmarking Out-Of-distribution Molecular Property Predictions of Machine Learning Models (NeurIPS 2025 · similarity 0.86)
- Pin-Tuning: Parameter-Efficient In-Context Tuning for Few-Shot Molecular Property Prediction (NeurIPS 2024 · similarity 0.84)
- Towards Foundational Models for Molecular Learning on Large-Scale Multi-Task Datasets (ICLR 2024 · similarity 0.84)
- Post Hoc Regression Refinement via Pairwise Rankings (NeurIPS 2025 · similarity 0.84)
- Learning Over Molecular Conformer Ensembles: Datasets and Benchmarks (ICLR 2024 · similarity 0.84)
- CheMixHub: Datasets and Benchmarks for Chemical Mixture Property Prediction (NeurIPS 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.
