Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning

Research output: Contribution to journal › Journal article › Research › peer-review

Standard

Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning. / Anker, Andy S.; Kjaer, Emil T. S.; Juelsholt, Mikkel; Christiansen, Troels Lindahl; Skjaervo, Susanne Linn; Jorgensen, Mads Ry Vogel; Kantor, Innokenty; Sorensen, Daniel Risskov; Billinge, Simon J. L.; Selvan, Raghavendra; Jensen, Kirsten M. O.

In: npj Computational Materials, Vol. 8, No. 1, 213, 01.10.2022.

Research output: Contribution to journal › Journal article › Research › peer-review

Harvard

Anker, AS, Kjaer, ETS, Juelsholt, M, Christiansen, TL, Skjaervo, SL, Jorgensen, MRV, Kantor, I, Sorensen, DR, Billinge, SJL, Selvan, R & Jensen, KMO 2022, 'Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning', npj Computational Materials, vol. 8, no. 1, 213. https://doi.org/10.1038/s41524-022-00896-3

APA

Anker, A. S., Kjaer, E. T. S., Juelsholt, M., Christiansen, T. L., Skjaervo, S. L., Jorgensen, M. R. V., Kantor, I., Sorensen, D. R., Billinge, S. J. L., Selvan, R., & Jensen, K. M. O. (2022). Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning. npj Computational Materials, 8(1), [213]. https://doi.org/10.1038/s41524-022-00896-3

Vancouver

Anker AS, Kjaer ETS, Juelsholt M, Christiansen TL, Skjaervo SL, Jorgensen MRV et al. Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning. npj Computational Materials. 2022 Oct 1;8(1). 213. https://doi.org/10.1038/s41524-022-00896-3

Author

Anker, Andy S. ; Kjaer, Emil T. S. ; Juelsholt, Mikkel ; Christiansen, Troels Lindahl ; Skjaervo, Susanne Linn ; Jorgensen, Mads Ry Vogel ; Kantor, Innokenty ; Sorensen, Daniel Risskov ; Billinge, Simon J. L. ; Selvan, Raghavendra ; Jensen, Kirsten M. O. / Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning. In: npj Computational Materials. 2022 ; Vol. 8, No. 1.

Bibtex

@article{04f3c90fd33f4f569bc10c3003f6ea29,
title = "Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning",
abstract = "Characterization of material structure with X-ray or neutron scattering using e.g. Pair Distribution Function (PDF) analysis most often rely on refining a structure model against an experimental dataset. However, identifying a suitable model is often a bottleneck. Recently, automated approaches have made it possible to test thousands of models for each dataset, but these methods are computationally expensive and analysing the output, i.e. extracting structural information from the resulting fits in a meaningful way, is challenging. Our Machine Learning based Motif Extractor (ML-MotEx) trains an ML algorithm on thousands of fits, and uses SHAP (SHapley Additive exPlanation) values to identify which model features are important for the fit quality. We use the method for 4 different chemical systems, including disordered nanomaterials and clusters. ML-MotEx opens for a type of modelling where each feature in a model is assigned an importance value for the fit quality based on explainable ML.",
keywords = "AB-INITIO DETERMINATION, ATOMIC-STRUCTURE, CRYSTAL, PROGRAM, CRYSTALLOGRAPHY, NANOPARTICLES, COMPLEX",
author = "Anker, {Andy S.} and Kjaer, {Emil T. S.} and Mikkel Juelsholt and Christiansen, {Troels Lindahl} and Skjaervo, {Susanne Linn} and Jorgensen, {Mads Ry Vogel} and Innokenty Kantor and Sorensen, {Daniel Risskov} and Billinge, {Simon J. L.} and Raghavendra Selvan and Jensen, {Kirsten M. O.}",
year = "2022",
month = oct,
day = "1",
doi = "10.1038/s41524-022-00896-3",
language = "English",
volume = "8",
journal = "npj Computational Materials",
issn = "2057-3960",
publisher = "Nature Publishing Group",
number = "1",

}

RIS

TY - JOUR

T1 - Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning

AU - Anker, Andy S.

AU - Kjaer, Emil T. S.

AU - Juelsholt, Mikkel

AU - Christiansen, Troels Lindahl

AU - Skjaervo, Susanne Linn

AU - Jorgensen, Mads Ry Vogel

AU - Kantor, Innokenty

AU - Sorensen, Daniel Risskov

AU - Billinge, Simon J. L.

AU - Selvan, Raghavendra

AU - Jensen, Kirsten M. O.

PY - 2022/10/1

Y1 - 2022/10/1

N2 - Characterization of material structure with X-ray or neutron scattering using e.g. Pair Distribution Function (PDF) analysis most often rely on refining a structure model against an experimental dataset. However, identifying a suitable model is often a bottleneck. Recently, automated approaches have made it possible to test thousands of models for each dataset, but these methods are computationally expensive and analysing the output, i.e. extracting structural information from the resulting fits in a meaningful way, is challenging. Our Machine Learning based Motif Extractor (ML-MotEx) trains an ML algorithm on thousands of fits, and uses SHAP (SHapley Additive exPlanation) values to identify which model features are important for the fit quality. We use the method for 4 different chemical systems, including disordered nanomaterials and clusters. ML-MotEx opens for a type of modelling where each feature in a model is assigned an importance value for the fit quality based on explainable ML.

AB - Characterization of material structure with X-ray or neutron scattering using e.g. Pair Distribution Function (PDF) analysis most often rely on refining a structure model against an experimental dataset. However, identifying a suitable model is often a bottleneck. Recently, automated approaches have made it possible to test thousands of models for each dataset, but these methods are computationally expensive and analysing the output, i.e. extracting structural information from the resulting fits in a meaningful way, is challenging. Our Machine Learning based Motif Extractor (ML-MotEx) trains an ML algorithm on thousands of fits, and uses SHAP (SHapley Additive exPlanation) values to identify which model features are important for the fit quality. We use the method for 4 different chemical systems, including disordered nanomaterials and clusters. ML-MotEx opens for a type of modelling where each feature in a model is assigned an importance value for the fit quality based on explainable ML.

KW - AB-INITIO DETERMINATION

KW - ATOMIC-STRUCTURE

KW - CRYSTAL

KW - PROGRAM

KW - CRYSTALLOGRAPHY

KW - NANOPARTICLES

KW - COMPLEX

U2 - 10.1038/s41524-022-00896-3

DO - 10.1038/s41524-022-00896-3

M3 - Journal article

VL - 8

JO - npj Computational Materials

JF - npj Computational Materials

SN - 2057-3960

IS - 1

M1 - 213

ER -

ID: 322198363