Improving direct physical properties prediction of heterogeneous materials from imaging data via convolutional neural network and a morphology-aware generative model

Ruijin Cang, Hechao Li, Hope Yao, Yang Jiao, Yi Ren

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Direct prediction of material properties from microstructures through statistical models has shown to be a potential approach to accelerating computational material design with large design spaces. However, statistical modeling of highly nonlinear mappings defined on high-dimensional microstructure spaces is known to be data-demanding. Thus, the added value of such predictive models diminishes in common cases where material samples (in forms of 2D or 3D microstructures) become costly to acquire either experimentally or computationally. To this end, we propose a generative machine learning model that creates an arbitrary amount of artificial material samples with negligible computation cost, when trained on only a limited amount of authentic samples. The key contribution of this work is the introduction of a morphology constraint to the training of the generative model, that enforces the resultant artificial material samples to have the same morphology distribution as the authentic ones. We show empirically that the proposed model creates artificial samples that better match with the authentic ones in material property distributions than those generated from a state-of-the-art Markov Random Field model, and thus is more effective at improving the prediction performance of a predictive structure-property model.

Original languageEnglish (US)
Pages (from-to)212-221
Number of pages10
JournalComputational Materials Science
Volume150
DOIs
StatePublished - Jul 1 2018

Fingerprint

Heterogeneous Materials
Generative Models
Physical property
Physical properties
physical properties
Imaging
Neural Networks
Neural networks
Imaging techniques
Prediction
Microstructure
predictions
Material Properties
Materials properties
microstructure
Material Design
Nonlinear Mapping
Statistical Modeling
Performance Prediction
Predictive Model

Keywords

  • Deep learning
  • Generative models
  • Integrated computational material engineering
  • Structure-property mapping

ASJC Scopus subject areas

  • Computer Science(all)
  • Chemistry(all)
  • Materials Science(all)
  • Mechanics of Materials
  • Physics and Astronomy(all)
  • Computational Mathematics

Cite this

@article{871deb03c04945109ac4ff8a67cb0275,
title = "Improving direct physical properties prediction of heterogeneous materials from imaging data via convolutional neural network and a morphology-aware generative model",
abstract = "Direct prediction of material properties from microstructures through statistical models has shown to be a potential approach to accelerating computational material design with large design spaces. However, statistical modeling of highly nonlinear mappings defined on high-dimensional microstructure spaces is known to be data-demanding. Thus, the added value of such predictive models diminishes in common cases where material samples (in forms of 2D or 3D microstructures) become costly to acquire either experimentally or computationally. To this end, we propose a generative machine learning model that creates an arbitrary amount of artificial material samples with negligible computation cost, when trained on only a limited amount of authentic samples. The key contribution of this work is the introduction of a morphology constraint to the training of the generative model, that enforces the resultant artificial material samples to have the same morphology distribution as the authentic ones. We show empirically that the proposed model creates artificial samples that better match with the authentic ones in material property distributions than those generated from a state-of-the-art Markov Random Field model, and thus is more effective at improving the prediction performance of a predictive structure-property model.",
keywords = "Deep learning, Generative models, Integrated computational material engineering, Structure-property mapping",
author = "Ruijin Cang and Hechao Li and Hope Yao and Yang Jiao and Yi Ren",
year = "2018",
month = "7",
day = "1",
doi = "10.1016/j.commatsci.2018.03.074",
language = "English (US)",
volume = "150",
pages = "212--221",
journal = "Computational Materials Science",
issn = "0927-0256",
publisher = "Elsevier",

}

TY - JOUR

T1 - Improving direct physical properties prediction of heterogeneous materials from imaging data via convolutional neural network and a morphology-aware generative model

AU - Cang, Ruijin

AU - Li, Hechao

AU - Yao, Hope

AU - Jiao, Yang

AU - Ren, Yi

PY - 2018/7/1

Y1 - 2018/7/1

N2 - Direct prediction of material properties from microstructures through statistical models has shown to be a potential approach to accelerating computational material design with large design spaces. However, statistical modeling of highly nonlinear mappings defined on high-dimensional microstructure spaces is known to be data-demanding. Thus, the added value of such predictive models diminishes in common cases where material samples (in forms of 2D or 3D microstructures) become costly to acquire either experimentally or computationally. To this end, we propose a generative machine learning model that creates an arbitrary amount of artificial material samples with negligible computation cost, when trained on only a limited amount of authentic samples. The key contribution of this work is the introduction of a morphology constraint to the training of the generative model, that enforces the resultant artificial material samples to have the same morphology distribution as the authentic ones. We show empirically that the proposed model creates artificial samples that better match with the authentic ones in material property distributions than those generated from a state-of-the-art Markov Random Field model, and thus is more effective at improving the prediction performance of a predictive structure-property model.

AB - Direct prediction of material properties from microstructures through statistical models has shown to be a potential approach to accelerating computational material design with large design spaces. However, statistical modeling of highly nonlinear mappings defined on high-dimensional microstructure spaces is known to be data-demanding. Thus, the added value of such predictive models diminishes in common cases where material samples (in forms of 2D or 3D microstructures) become costly to acquire either experimentally or computationally. To this end, we propose a generative machine learning model that creates an arbitrary amount of artificial material samples with negligible computation cost, when trained on only a limited amount of authentic samples. The key contribution of this work is the introduction of a morphology constraint to the training of the generative model, that enforces the resultant artificial material samples to have the same morphology distribution as the authentic ones. We show empirically that the proposed model creates artificial samples that better match with the authentic ones in material property distributions than those generated from a state-of-the-art Markov Random Field model, and thus is more effective at improving the prediction performance of a predictive structure-property model.

KW - Deep learning

KW - Generative models

KW - Integrated computational material engineering

KW - Structure-property mapping

UR - http://www.scopus.com/inward/record.url?scp=85045220619&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85045220619&partnerID=8YFLogxK

U2 - 10.1016/j.commatsci.2018.03.074

DO - 10.1016/j.commatsci.2018.03.074

M3 - Article

VL - 150

SP - 212

EP - 221

JO - Computational Materials Science

JF - Computational Materials Science

SN - 0927-0256

ER -