skip to main content

DOE PAGESDOE PAGES

Title: Inferring low-dimensional microstructure representations using convolutional neural networks

We apply here recent advances in machine learning and computer vision to a central problem in materials informatics: the statistical representation of microstructural images. We use activations in a pretrained convolutional neural network to provide a high-dimensional characterization of a set of synthetic microstructural images. Next, we use manifold learning to obtain a low-dimensional embedding of this statistical characterization. We show that the low-dimensional embedding extracts the parameters used to generate the images. Finally, according to a variety of metrics, the convolutional neural network method yields dramatically better embeddings than the analogous method derived from two-point correlations alone.
Authors:
 [1] ;  [2] ;  [2]
  1. Boston Univ., MA (United States). Dept. of Physics; Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
  2. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Publication Date:
Report Number(s):
LA-UR-16-27229
Journal ID: ISSN 2470-0045
Grant/Contract Number:
AC52-06NA25396
Type:
Accepted Manuscript
Journal Name:
Physical Review E
Additional Journal Information:
Journal Volume: 96; Journal Issue: 5; Journal ID: ISSN 2470-0045
Publisher:
American Physical Society (APS)
Research Org:
Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Sponsoring Org:
USDOE; LANL Laboratory Directed Research and Development (LDRD) Program
Country of Publication:
United States
Language:
English
Subject:
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSICS; microstructure; coarse graining; machine learning
OSTI Identifier:
1485399
Alternate Identifier(s):
OSTI ID: 1408180

Lubbers, Nicholas, Lookman, Turab, and Barros, Kipton. Inferring low-dimensional microstructure representations using convolutional neural networks. United States: N. p., Web. doi:10.1103/PhysRevE.96.052111.
Lubbers, Nicholas, Lookman, Turab, & Barros, Kipton. Inferring low-dimensional microstructure representations using convolutional neural networks. United States. doi:10.1103/PhysRevE.96.052111.
Lubbers, Nicholas, Lookman, Turab, and Barros, Kipton. 2017. "Inferring low-dimensional microstructure representations using convolutional neural networks". United States. doi:10.1103/PhysRevE.96.052111. https://www.osti.gov/servlets/purl/1485399.
@article{osti_1485399,
title = {Inferring low-dimensional microstructure representations using convolutional neural networks},
author = {Lubbers, Nicholas and Lookman, Turab and Barros, Kipton},
abstractNote = {We apply here recent advances in machine learning and computer vision to a central problem in materials informatics: the statistical representation of microstructural images. We use activations in a pretrained convolutional neural network to provide a high-dimensional characterization of a set of synthetic microstructural images. Next, we use manifold learning to obtain a low-dimensional embedding of this statistical characterization. We show that the low-dimensional embedding extracts the parameters used to generate the images. Finally, according to a variety of metrics, the convolutional neural network method yields dramatically better embeddings than the analogous method derived from two-point correlations alone.},
doi = {10.1103/PhysRevE.96.052111},
journal = {Physical Review E},
number = 5,
volume = 96,
place = {United States},
year = {2017},
month = {11}
}