Evaluation of machine learning interpolation techniques for prediction of physical properties

Belisle, Eve, Huang, Zi, Le Digabel, Sebastien and Gheribi, Aimen E. (2015) Evaluation of machine learning interpolation techniques for prediction of physical properties. Computational Materials Science, 98 170-177. doi:10.1016/j.commatsci.2014.10.032


Author Belisle, Eve
Huang, Zi
Le Digabel, Sebastien
Gheribi, Aimen E.
Title Evaluation of machine learning interpolation techniques for prediction of physical properties
Journal name Computational Materials Science   Check publisher's open access policy
ISSN 0927-0256
Publication date 2015-02-15
Year available 2014
Sub-type Article (original research)
DOI 10.1016/j.commatsci.2014.10.032
Open Access Status
Volume 98
Start page 170
End page 177
Total pages 8
Place of publication Amsterdam, Netherlands
Publisher Elsevier
Collection year 2015
Language eng
Formatted abstract
A knowledge of the physical properties of materials as a function of temperature, composition, applied external stresses, etc. is an important consideration in materials and process design. For new systems, such properties may be unknown and hard to measure or estimate from numerical simulations such as molecular dynamics. Engineers rely on machine learning to employ existing data in order to predict properties for new systems. Several techniques are currently used for such purposes. These include neural network, polynomial interpolation and Gaussian processes as well as the more recent dynamic trees and scalable Gaussian processes. In this paper we compare these approaches for three sets of materials sciences data: molar volume, electrical conductivity and Martensite start temperature. We make recommendations depending on the nature of the data. We demonstrate that a thorough knowledge of the problem beforehand is critical in selecting the most successful machine learning technique. Our findings show that the Gaussian process regression technique gives very good predictions for all three sets of tested data. Typically, Gaussian process is very slow with a computational complexity of typically n3 where n is the number of data points. In this paper, we found that the scalable Gaussian process approach was able to maintain the high accuracy of the predictions while improving speed considerably, make on-line learning possible.
Keyword Superalloys
Database
Gaussian process
Neural network
Quadratic regression
Physical properties
Computational dependence
Q-Index Code C1
Q-Index Status Confirmed Code
Institutional Status UQ
Additional Notes Published online ahead of print 27 Nov 2014

Document type: Journal Article
Sub-type: Article (original research)
Collections: Official 2015 Collection
School of Information Technology and Electrical Engineering Publications
 
Versions
Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 5 times in Thomson Reuters Web of Science Article | Citations
Scopus Citation Count Cited 5 times in Scopus Article | Citations
Google Scholar Search Google Scholar
Created: Tue, 16 Dec 2014, 00:14:38 EST by System User on behalf of School of Information Technol and Elec Engineering