Focused Session CB-7
Quantification of Microstructures using Data Analysis and Machine Learning Methods

Kwang-Ryeol LEE, KIST, South Korea
Veena TIKARE, Sandia National Laboratories, USA (Programme Chair)
Jui-Sheng CHOU, Taiwan Tech, Taiwan
Masahiko DEMURA, NIMS, Japan
Raynald GAUVIN, McGill University, Canada
Satoshi HATA, Kyushu University, Japan
Elizabeth HOLM, Carnegie Mellon University, USA
Surya KALIDINDI, Georgia Institute of Technology, USA
Bryce MEREDIG, Citrine Informatics, USA
Veera SUNDARARAGHAVAN, University of Michigan, Ann Arbor, USA
Ichiro TAKEUCHI, University of Maryland, USA
Francois WILLOT, MINES ParisTech, France
Aron WALSH, Imperial College London, UK
Chris, WOLVERTON, Northwestern University, USA
Leonardo AGUDO JÁCOME, BAM, Germany
Hamish L. FRASER, Ohio State University, USA
Sang Soo HAN, Korea Institute of Science and Technology, South Korea
Elizabeth A. HOLM, Carnegie Mellon University, USA
Surya R. KALIDINDI, Georgia Institute of Technology, USA
Boris KOZINSKY, Harvard University, USA
Jehyun LEE, Korea Institute of Energy Research, South Korea
Shunsuke MUTO, Nagoya University, Japan
Stefanos PAPANIKOLAOU, West Virginia University, USA
Giovanni PIZZI, EPFL, Switzerland
Volker SCHMIDT, University of Ulm, Germany
Thomas SCHREFL, Danube University Krems, Austria
Ichiro TAKEUCHI, University of Maryland, USA
Chris WOLVERTON, Northwestern University, USA
Central to materials science are the links between processing and microstructure, and between microstructure and properties. Advances made in recent years include the ability to image microstructures over large areas with high resolution using multiple electron beams that raster simultaneously in SEM and obtaining high resolution 3D images using tomography or reconstruction methods. However, quantitative characterization of these microstructures is still rudimentary, which in turns limits finding quantitative relationships between processing and microstructure, and microstructure and properties. Developing methods to quantify size, shape, curvature, texture, distributions of these quantities, topology, connectivity, and spatial correlations between such characteristics of microstructure features would greatly enhance our ability to design and fabricate ceramics microstructures to optimize properties. This symposium will focus on techniques to quantifying microstructure and its variation. We seek contributions on traditional methods based on stereology as well as newer methods using data analysis and machine learning that will advance our ability to rigorously quantify and compare the full range of microstructures observed in materials from dense uniform microstructures to highly varied ones with pores, intra-and inter-granular precipitates, grain boundary phases, additional solid phases, and compositional and other gradients. In addition to characterizing features at microstructural scale, methods that describe spatial correlation or hierarchical relationships of microstructural features at longer length scales are also solicited.

Matter of interest:
  • Automated application of traditional stereological techniques and their extension to three-dimensional microstructures.
  • Application of data analysis techniques such as multi-point statistics, primary component analysis, object-based image analysis, feature identification and extraction, spatial correlation and 3D pose estimation.
  • Use of machine learning to identify key features and the correct metrics to quantify their characteristics, find spatial correlations or hierarchical relations and other relationships that are not possible when microstructures are analyzed manually.
  • While the focus is on microstructure quantification, works that relate microstructure to either fabrication processes or to engineering properties using data analytics or machine learning will be considered.
Session Topics

CB-7.1 Automation of stereological techniques to characterize microstructure

CB-7.2 Data analysis for quantitative description of microstructure (i.e. multi-point methods, primary component analysis, spectral methods, etc.)

CB-7.3 Machine learning to recognize microstructural features and quantify microstructures

CB-7.4 Development of long-range descriptors (i.e. spatial correlation, hierarchical relations)

CB-7.5 Application of data analytics or machine learning to find processing-microstructure or microstructure-property relationships


Cimtec 2021

Copyright © Techna Group S.r.l.
C.F.-P.I. 03368230409
Privacy Policy - Cookie Policy - Software Commercio Elettronico by Pianetaitalia.com