Ensemble Knowledge Distillation for Machine Learning Interatomic Potentials
Sakib Matin, Emily Shinkle, Yulia Pimonova, Galen T. Craven, Aleksandra Pachalieva, Ying Wai Li, Kipton Barros, Nicholas Lubbers · 2025
Read Ensemble Knowledge Distillation for Machine Learning Interatomic Potentials by Sakib Matin, Emily Shinkle, Yulia Pimonova, Galen T. Craven, Aleksandra Pachalieva, Ying Wai Li, Kipton Barros, Nicholas Lubbers (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-14293.txt.
Description
Sakib Matin and co-authors improve machine learning potentials for atoms by teaching a student model with a team of teachers. Their twenty twenty-five paper tackles a data gap in quantum chemistry. High-grade energy data often cover only small molecules and lack force values, which training needs.
Several teacher models first learn the energies, then supply forces for every shape in the set. The student learns both the true energies and the averaged teacher forces. Tests on a coupled-cluster data set show clear accuracy gains. Better potentials mean faster, truer simulations of how molecules move and react, without paying full quantum cost each step.
Catalog record
| CREATOR | Sakib Matin, Emily Shinkle, Yulia Pimonova, Galen T. Craven, Aleksandra Pachalieva, Ying Wai Li, Kipton Barros, Nicholas Lubbers |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-14293