Out-of-distribution evaluation of active learning pipelines for molecular property prediction
Open Access
- 21 January 2026
- journal article
- research article
- Published by Royal Society of Chemistry (RSC) in RSC Advances
- Vol. 16 (6) , 5281-5295
Abstract
Evidential deep learning–driven active learning beats random sampling for OOD solvation free-energy prediction, delivering lower test RMSE at the same labeling budget and improving generalization to unseen chemical space.Keywords
Funding Information
- Pacific Northwest National Laboratory
- Battelle (DE-AC05-76RL01830)
This publication has 27 references indexed in Scilit:
- Known Unknowns: Out-of-Distribution Property Prediction in Materials and Moleculesnpj Computational Materials, 2025
- Evaluation of Deep Learning Architectures for Aqueous Solubility PredictionACS Omega, 2022
- Solvation Free Energies for Aqueous and Nonaqueous Solutions Computed Using PM7 Atomic ChargesJournal of Chemical Information and Modeling, 2021
- Evidential Deep Learning for Guided Molecular Property Prediction and DiscoveryACS Central Science, 2021
- Ensemble learning-iterative training machine learning for uncertainty quantification and automated experiment in atom-resolved microscopynpj Computational Materials, 2021
- Fast non‐iterative calculation of solvation energies for water and non‐aqueous solventsJournal of Computational Chemistry, 2021
- GFN2-xTB—An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion ContributionsJournal of Chemical Theory and Computation, 2019
- Fréchet ChemNet Distance: A Metric for Generative Models for Molecules in Drug DiscoveryJournal of Chemical Information and Modeling, 2018
- Optimization of parameters for semiempirical methods VI: more modifications to the NDDO approximations and re-optimization of parametersJournal of Molecular Modeling, 2012
- Optimization of parameters for semiempirical methods V: Modification of NDDO approximations and application to 70 elementsJournal of Molecular Modeling, 2007