Compact representations of microstructure images using triplet networks
- Author
- Michiel Larmuseau, Michael Sluydts (UGent) , Koenraad Theuwissen, Lode Duprez, Tom Dhaene (UGent) and Stefaan Cottenier (UGent)
- Organization
- Abstract
- The microstructure of a material, typically characterized through a set of microscopy images of two-dimensional cross-sections, is a valuable source of information about the material and its properties. Every pixel of the image is a degree of freedom causing the dimensionality of the information space to be extremely high. This makes it difficult to recognize and extract all relevant information from the images. Human experts circumvent this by manually creating a lower-dimensional representation of the microstructure. However, the question of how a microstructure image can be best represented remains open. From the field of deep learning, we present triplet networks as a method to build highly compact representations of the microstructure, condensing the relevant information into a much smaller number of dimensions. We demonstrate that these representations can be created even with a limited amount of example images, and that they are able to distinguish between visually very similar microstructures. We discuss the interpretability and generalization of the representations. Having compact microstructure representations, it becomes easier to establish processing-structure-property links that are key to rational materials design.
- Keywords
- CLASSIFICATION, FEATURES
Downloads
-
7796.pdf
- full text (Published version)
- |
- open access
- |
- |
- 2.36 MB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-8679269
- MLA
- Larmuseau, Michiel, et al. “Compact Representations of Microstructure Images Using Triplet Networks.” NPJ COMPUTATIONAL MATERIALS, vol. 6, no. 1, 2020, doi:10.1038/s41524-020-00423-2.
- APA
- Larmuseau, M., Sluydts, M., Theuwissen, K., Duprez, L., Dhaene, T., & Cottenier, S. (2020). Compact representations of microstructure images using triplet networks. NPJ COMPUTATIONAL MATERIALS, 6(1). https://doi.org/10.1038/s41524-020-00423-2
- Chicago author-date
- Larmuseau, Michiel, Michael Sluydts, Koenraad Theuwissen, Lode Duprez, Tom Dhaene, and Stefaan Cottenier. 2020. “Compact Representations of Microstructure Images Using Triplet Networks.” NPJ COMPUTATIONAL MATERIALS 6 (1). https://doi.org/10.1038/s41524-020-00423-2.
- Chicago author-date (all authors)
- Larmuseau, Michiel, Michael Sluydts, Koenraad Theuwissen, Lode Duprez, Tom Dhaene, and Stefaan Cottenier. 2020. “Compact Representations of Microstructure Images Using Triplet Networks.” NPJ COMPUTATIONAL MATERIALS 6 (1). doi:10.1038/s41524-020-00423-2.
- Vancouver
- 1.Larmuseau M, Sluydts M, Theuwissen K, Duprez L, Dhaene T, Cottenier S. Compact representations of microstructure images using triplet networks. NPJ COMPUTATIONAL MATERIALS. 2020;6(1).
- IEEE
- [1]M. Larmuseau, M. Sluydts, K. Theuwissen, L. Duprez, T. Dhaene, and S. Cottenier, “Compact representations of microstructure images using triplet networks,” NPJ COMPUTATIONAL MATERIALS, vol. 6, no. 1, 2020.
@article{8679269,
abstract = {{The microstructure of a material, typically characterized through a set of microscopy images of two-dimensional cross-sections, is a valuable source of information about the material and its properties. Every pixel of the image is a degree of freedom causing the dimensionality of the information space to be extremely high. This makes it difficult to recognize and extract all relevant information from the images. Human experts circumvent this by manually creating a lower-dimensional representation of the microstructure. However, the question of how a microstructure image can be best represented remains open. From the field of deep learning, we present triplet networks as a method to build highly compact representations of the microstructure, condensing the relevant information into a much smaller number of dimensions. We demonstrate that these representations can be created even with a limited amount of example images, and that they are able to distinguish between visually very similar microstructures. We discuss the interpretability and generalization of the representations. Having compact microstructure representations, it becomes easier to establish processing-structure-property links that are key to rational materials design.}},
articleno = {{156}},
author = {{Larmuseau, Michiel and Sluydts, Michael and Theuwissen, Koenraad and Duprez, Lode and Dhaene, Tom and Cottenier, Stefaan}},
issn = {{2057-3960}},
journal = {{NPJ COMPUTATIONAL MATERIALS}},
keywords = {{CLASSIFICATION,FEATURES}},
language = {{eng}},
number = {{1}},
pages = {{11}},
title = {{Compact representations of microstructure images using triplet networks}},
url = {{http://doi.org/10.1038/s41524-020-00423-2}},
volume = {{6}},
year = {{2020}},
}
- Altmetric
- View in Altmetric
- Web of Science
- Times cited: