Smarter Data: Rethinking Data Generation for Machine Learning Potentials in Heterogeneous Catalysis
Wenbo Xie, Yixiao Han, Chenyu Wu, P. Hu · 2026
Read Smarter Data: Rethinking Data Generation for Machine Learning Potentials in Heterogeneous Catalysis by Wenbo Xie, Yixiao Han, Chenyu Wu, P. Hu (2026) 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/pmc-13291977.txt.
Description
Wenbo Xie and co-authors argue that training data choice, not model design, limits machine learning potentials in catalysis, in 2026. These potentials mimic quantum accuracy at far larger scales, promising realistic catalyst simulations. The catch is that a model only interpolates where its training examples have been. Catalyst surfaces shift shape, coverage, and chemistry as reactions run, widening the needed domain. The perspective sorts current data-building strategies by scope, relevance, and coverage.
Structure-based sampling, the prevailing habit, bakes relevance in by picking configurations from target systems. Interaction-based sampling tries instead to widen the local chemical variety a model sees. Each style leaves gaps the other fills, the authors contend. They close by sketching a hybrid workflow that blends both sampling logics. Their plea is to design datasets deliberately rather than simply grow them. Smarter data, they argue, is the shortest path to trustworthy catalytic potentials.
Catalog record
| CREATOR | Wenbo Xie, Yixiao Han, Chenyu Wu, P. Hu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |