Student or educator? Sign up with your school email and get 6 months of Pro free.

The DOI matches its Crossref record; 60 of the 60 references it cites were found in a registry or an index.

Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks

Open access

Abstract

X-ray diffraction (XRD) data acquisition and analysis is among the most time-consuming steps in the development cycle of novel thin-film materials. We propose a machine learning-enabled approach to predict crystallographic dimensionality and space group from a limited number of thin-film XRD patterns. We overcome the scarce data problem intrinsic to novel materials development by coupling a supervised machine learning approach with a model-agnostic, physics-informed data augmentation strategy using simulated data from the Inorganic Crystal Structure Database (ICSD) and experimental data. As a test case, 115 thin-film metal-halides spanning three dimensionalities and seven space groups are synthesized and classified. After testing various algorithms, we develop and implement an all convolutional neural network, with cross-validated accuracies for dimensionality and space group classification of 93 and 89%, respectively. We propose average class activation maps, computed from a global average pooling layer, to allow high model interpretability by human experimentalists, elucidating the root causes of misclassification. Finally, we systematically evaluate the maximum XRD pattern step size (data acquisition rate) before loss of predictive accuracy occurs, and determine it to be 0.16° 2 θ , which enables an XRD pattern to be obtained and classified in 5.5 min or less.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41524-019-0196-x

Cites

60 of 60 references verified0 citation errors · 0 not checkable

Statistics

Figures

Image 1
CC-BY · From the PDF
Fig. 1 General framework and a-CNN architecture. a Schematic of our X-ray diffraction data classi fi cation framework, with physics-informed data augmentation. b Schematic of the best-performing algorithm, our all convolutional neural network
Fig. 1 General framework and a-CNN architecture. a Schematic of our X-ray diffraction data classi fi cation framework, with physics-informed data augmentation. b Schematic of the best-performing algorithm, our all convolutional neural networkCC-BY · From the PDF
Fig. 2 Physics-informed data augmentation: schematic of the physics-informed data augmentation algorithm, which accounts for the particularities of thin- fi lm XRD spectra, as described by the subsequent transformations stated in Eqs. 1 -3
Fig. 2 Physics-informed data augmentation: schematic of the physics-informed data augmentation algorithm, which accounts for the particularities of thin- fi lm XRD spectra, as described by the subsequent transformations stated in Eqs. 1 -3CC-BY · From the PDF
Image 4
CC-BY · From the PDF
Fig. 3 Accuracy of the a-CNN model with augmented data: line plot showing mean Case 3 a-CNN accuracy, as a function of the number of augmented patterns (Methods, Eqs. 1 -3) included in the training set, for ( a ) dimensionality classi fi cation and ( b ) space group classi fi cation. The x axis shows augmented experimental data (based on the original experimental XRD patterns), and the legend shows simulated data (based on the 164 simulated powder diffraction patterns obtained from the ICSD). The error bars correspond to one standard deviation from the mean
Fig. 3 Accuracy of the a-CNN model with augmented data: line plot showing mean Case 3 a-CNN accuracy, as a function of the number of augmented patterns (Methods, Eqs. 1 -3) included in the training set, for ( a ) dimensionality classi fi cation and ( b ) space group classi fi cation. The x axis shows augmented experimental data (based on the original experimental XRD patterns), and the legend shows simulated data (based on the 164 simulated powder diffraction patterns obtained from the ICSD). The error bars correspond to one standard deviation from the meanCC-BY · From the PDF
Fig. 4 Reducing XRD pattern acquisition time: simulation of the trade-off between XRD pattern acquisition rate and predictive accuracy. Accuracies for crystal dimensionality and space group predictions are estimated by coarsening the XRD spectrum 2 ϴ step size for Case 3 conditions. The error bars correspond to one standard deviation from the mean
Fig. 4 Reducing XRD pattern acquisition time: simulation of the trade-off between XRD pattern acquisition rate and predictive accuracy. Accuracies for crystal dimensionality and space group predictions are estimated by coarsening the XRD spectrum 2 ϴ step size for Case 3 conditions. The error bars correspond to one standard deviation from the meanCC-BY · From the PDF
Fig. 5 Class activation maps for misclassi fi cation interpretability: class activation maps (CAMs) generated by the a-CNN architecture, representing space group classi fi cation of Class 2 ( P 21/ a ) and Class 6 ( Pm 3 m ). ( a , b ) correspond to maps generated by averaging all training samples in a certain class, whereas ( c , d ) correspond to the CAMs of correctly classi fi ed and incorrectly classi fi ed individual patterns, respectively. The correctly predicted pattern of ( c ) is explained by the similarity of its CAM to the average CAM of Class 6; whereas, the incorrectly prediction ( d ) can be explained by comparing its CAM with the average CAMs of classes 2 and 6 5
Fig. 5 Class activation maps for misclassi fi cation interpretability: class activation maps (CAMs) generated by the a-CNN architecture, representing space group classi fi cation of Class 2 ( P 21/ a ) and Class 6 ( Pm 3 m ). ( a , b ) correspond to maps generated by averaging all training samples in a certain class, whereas ( c , d ) correspond to the CAMs of correctly classi fi ed and incorrectly classi fi ed individual patterns, respectively. The correctly predicted pattern of ( c ) is explained by the similarity of its CAM to the average CAM of Class 6; whereas, the incorrectly prediction ( d ) can be explained by comparing its CAM with the average CAMs of classes 2 and 6 5CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 57 references

Research works printed but not linked: 2

  1. Goodfellow, I., Bengio, Y., Courville, A. & Bengio, Y. Deep Learning. 1 (MIT Press, Cambridge, 2016).
  2. Kobayashi, S. & Inaba, K. X-ray thin-film measurement techniques. Mass Spectrosc. Equipped a Ski. Interface 28, 8 (2012).

Cited by

See all 327

Metadata

Authors
Oviedo, F., Ren, Z., Sun, S., Settens, C., Liu, Z., Hartono, N. T. P., Ramasamy, S., DeCost, B. L., Tian, S. I. P., Romano, G., Gilad Kusne, A., & Buonassisi, T.
Affiliations
  • Massachusetts Institute of Technology (US)
  • Singapore-MIT Alliance for Research and Technology (SG)
  • Agency for Science, Technology and Research (SG)
  • National Institute of Standards and Technology (US)
Corresponding author
  • Oviedo, Felipe
  • Ren, Zekun
  • Buonassisi, Tonio
Venue
npj Computational Materials
Published
2019-05-17
Volume
5
Issue
1
ISSN
2057-3960 (ISSN-L)
OpenAlex
W2940802411
Cited by
327
References
57
Citations by year
2026: 36 · 2025: 50 · 2024: 49 · 2023: 47 · 2022: 56 · 2021: 53 · 2020: 30 · 2019: 6
FWCI
13.09
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Computational Materials (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
  • arXiv (Cornell University)
Funders
  • National Science Foundation (0819762)
  • U.S. Department of Energy (DEEE0007535)
  • Massachusetts Institute of Technology
  • National Research Foundation
  • Agency for Science, Technology and Research (A1898b0043)
  • National Institute of Standards and Technology
  • Division of Materials Research (DMR - 0819762, 0819762)
Topics
  • Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
  • Physical Sciences › Materials Science › Materials Chemistry › X-ray Diffraction in Crystallography
  • Physical Sciences › Chemistry › Inorganic Chemistry › Crystal structures of chemical compounds
Keywords
thin films, metal halides, data augmentation, physics-informed data augmentation, simulated diffraction patterns, Inorganic Crystal Structure Database, convolutional neural networks, class activation maps, model interpretability, data acquisition rate, data scarcity

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks. https://app.cochat.ai/papers/10.1038/s41524-019-0196-x
The paper
Oviedo, F., Ren, Z., Sun, S., Settens, C., Liu, Z., Hartono, N. T. P., Ramasamy, S., DeCost, B. L., Tian, S. I. P., Romano, G., Gilad Kusne, A., & Buonassisi, T. (2019). Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks. npj Computational Materials. https://doi.org/10.1038/s41524-019-0196-x

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.