Unsupervised machine learning discovery of structural units and transformation pathways from imaging data
We show that unsupervised machine learning can be used to learn chemical transformation pathways from observational Scanning Transmission Electron Microscopy (STEM) data. To enable this analysis, we assumed the existence of atoms, a discreteness of atomic classes, and the presence of an explicit rel...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
AIP Publishing LLC
2023-06-01
|
| Series: | APL Machine Learning |
| Online Access: | http://dx.doi.org/10.1063/5.0147316 |
| _version_ | 1797635551576195072 |
|---|---|
| author | Sergei V. Kalinin Ondrej Dyck Ayana Ghosh Yongtao Liu Bobby G. Sumpter Maxim Ziatdinov |
| author_facet | Sergei V. Kalinin Ondrej Dyck Ayana Ghosh Yongtao Liu Bobby G. Sumpter Maxim Ziatdinov |
| author_sort | Sergei V. Kalinin |
| collection | DOAJ |
| description | We show that unsupervised machine learning can be used to learn chemical transformation pathways from observational Scanning Transmission Electron Microscopy (STEM) data. To enable this analysis, we assumed the existence of atoms, a discreteness of atomic classes, and the presence of an explicit relationship between the observed STEM contrast and the presence of atomic units. With only these postulates, we developed a machine learning method leveraging a rotationally invariant variational autoencoder (VAE) that can identify the existing molecular fragments observed within a material. The approach encodes the information contained in STEM image sequences using a small number of latent variables, allowing the exploration of chemical transformation pathways by tracing the evolution of atoms in the latent space of the system. The results suggest that atomically resolved STEM data can be used to derive fundamental physical and chemical mechanisms involved, by providing encodings of the observed structures that act as bottom-up equivalents of structural order parameters. The approach also demonstrates the potential of variational (i.e., Bayesian) methods in the physical sciences and will stimulate the development of more sophisticated ways to encode physical constraints in the encoder–decoder architectures and generative physical laws and causal relationships in the latent space of VAEs. |
| first_indexed | 2024-03-11T12:22:22Z |
| format | Article |
| id | doaj.art-057f9671de0f4b9e9e07456e9326c131 |
| institution | Directory Open Access Journal |
| issn | 2770-9019 |
| language | English |
| last_indexed | 2024-03-11T12:22:22Z |
| publishDate | 2023-06-01 |
| publisher | AIP Publishing LLC |
| record_format | Article |
| series | APL Machine Learning |
| spelling | doaj.art-057f9671de0f4b9e9e07456e9326c1312023-11-06T21:03:50ZengAIP Publishing LLCAPL Machine Learning2770-90192023-06-0112026117026117-1010.1063/5.0147316Unsupervised machine learning discovery of structural units and transformation pathways from imaging dataSergei V. Kalinin0Ondrej Dyck1Ayana Ghosh2Yongtao Liu3Bobby G. Sumpter4Maxim Ziatdinov5Department of Materials Science and Engineering, The University of Tennessee, Knoxville, Tennessee 37996, USACenter for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USAComputational Sciences and Engineering Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USACenter for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USACenter for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USACenter for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USAWe show that unsupervised machine learning can be used to learn chemical transformation pathways from observational Scanning Transmission Electron Microscopy (STEM) data. To enable this analysis, we assumed the existence of atoms, a discreteness of atomic classes, and the presence of an explicit relationship between the observed STEM contrast and the presence of atomic units. With only these postulates, we developed a machine learning method leveraging a rotationally invariant variational autoencoder (VAE) that can identify the existing molecular fragments observed within a material. The approach encodes the information contained in STEM image sequences using a small number of latent variables, allowing the exploration of chemical transformation pathways by tracing the evolution of atoms in the latent space of the system. The results suggest that atomically resolved STEM data can be used to derive fundamental physical and chemical mechanisms involved, by providing encodings of the observed structures that act as bottom-up equivalents of structural order parameters. The approach also demonstrates the potential of variational (i.e., Bayesian) methods in the physical sciences and will stimulate the development of more sophisticated ways to encode physical constraints in the encoder–decoder architectures and generative physical laws and causal relationships in the latent space of VAEs.http://dx.doi.org/10.1063/5.0147316 |
| spellingShingle | Sergei V. Kalinin Ondrej Dyck Ayana Ghosh Yongtao Liu Bobby G. Sumpter Maxim Ziatdinov Unsupervised machine learning discovery of structural units and transformation pathways from imaging data APL Machine Learning |
| title | Unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| title_full | Unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| title_fullStr | Unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| title_full_unstemmed | Unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| title_short | Unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| title_sort | unsupervised machine learning discovery of structural units and transformation pathways from imaging data |
| url | http://dx.doi.org/10.1063/5.0147316 |
| work_keys_str_mv | AT sergeivkalinin unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata AT ondrejdyck unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata AT ayanaghosh unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata AT yongtaoliu unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata AT bobbygsumpter unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata AT maximziatdinov unsupervisedmachinelearningdiscoveryofstructuralunitsandtransformationpathwaysfromimagingdata |