Model-free estimation of completeness, uncertainties, and outliers in atomistic machine learning using information theory

Journal Article · 28 April 2025 · Nature Communications

DOI: https://doi.org/10.1038/s41467-025-59232-0 · OSTI ID:2563460

Schwalbe-Koda, Daniel ; Hamel, Sebastien ; Sadigh, Babak; Zhou, Fei ; Lordi, Vincenzo

Abstract An accurate description of information is relevant for a range of problems in atomistic machine learning (ML), such as crafting training sets, performing uncertainty quantification (UQ), or extracting physical insights from large datasets. However, atomistic ML often relies on unsupervised learning or model predictions to analyze information contents from simulation or training data. Here, we introduce a theoretical framework that provides a rigorous, model-free tool to quantify information contents in atomistic simulations. We demonstrate that the information entropy of a distribution of atom-centered environments explains known heuristics in ML potential developments, from training set sizes to dataset optimality. Using this tool, we propose a model-free UQ method that reliably predicts epistemic uncertainty and detects out-of-distribution samples, including rare events in systems such as nucleation. This method provides a general tool for data-driven atomistic modeling and combines efforts in ML, simulations, and physical explainability.

• Research Organization: Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States); University of California, Los Angeles, CA (United States)
• Sponsoring Organization: USDOE; USDOE Laboratory Directed Research and Development (LDRD) Program; USDOE Office of Science (SC), Basic Energy Sciences (BES)
• Grant/Contract Number: AC52-07NA27344; SC0025642
• OSTI ID: 2563460
• Report Number(s): LLNL--JRNL-862887; LLNL--LDRD 22-ERD-055; LLNL--LDRD 23-SI-006; 4014; PII: 59232
• Journal Information: Nature Communications, Journal Name: Nature Communications Journal Issue: 1 Vol. 16; ISSN 2041-1723
• Publisher: Nature Publishing GroupCopyright Statement
• Country of Publication: United Kingdom
• Language: English