header

Semi-Supervised Learning Approaches to Class Assignment in Ambiguous Microstructures

26 Pages Posted: 10 Dec 2019 Publication Status: Accepted

See all articles by Courtney Kunselman

Courtney Kunselman

Texas A&M University - Department of Materials Science and Engineering

Vahid Attari

Texas A&M University - Department of Materials Science and Engineering

Levi McClenny

Texas A&M University - Department of Electrical Engineering

Ulisses Braga-Neto

Texas A&M University - Department of Electrical Engineering

Raymundo Arroyave

Texas A&M University - Department of Materials Science and Engineering; Texas A&M University - Center for intelligent Multifunctional Materials and Structures; Texas A&M University - Department of Mechanical Engineering

Abstract

Uncovering links between processing conditions, microstructure, and properties is a central tenet of materials analysis. It is well known that microstructure determines properties, but expressing these structural features in a universal quantitative fashion has proved to be extremely difficult. Recent efforts have focused on training supervised learning algorithms to place microstructure images into predefined classes, but this approach assumes a level of a priori knowledge which is not always available. In this paper, we expand this idea to the semi-supervised context in which class labels are known with confidence for only a fraction of the microstructures that represent the material system. It is shown that classifiers which perform well on both the high-confidence labeled data and the unlabeled, ambiguous data can be constructed by relying on the labeling consensus of a collection of semi-supervised learning methods. We also demonstrate the use of novel error estimation approaches for unlabeled data to establish robust confidence bounds on the classification performance over the entire microstructure space.

Keywords: Machine learning, Microstructure classification, Support vector machines, Semi-supervised learning methods, Unsupervised error estimation

Suggested Citation

Kunselman, Courtney and Attari, Vahid and McClenny, Levi and Braga-Neto, Ulisses and Arroyave, Raymundo, Semi-Supervised Learning Approaches to Class Assignment in Ambiguous Microstructures (December 5, 2019). Available at SSRN: https://ssrn.com/abstract=3499091 or http://dx.doi.org/10.2139/ssrn.3499091

Courtney Kunselman (Contact Author)

Texas A&M University - Department of Materials Science and Engineering ( email )

Langford Building A
798 Ross St.
College Station, TX 77843-3137
United States

Vahid Attari

Texas A&M University - Department of Materials Science and Engineering ( email )

Langford Building A
798 Ross St.
College Station, TX 77843-3137
United States

Levi McClenny

Texas A&M University - Department of Electrical Engineering

United States

Ulisses Braga-Neto

Texas A&M University - Department of Electrical Engineering

United States

Raymundo Arroyave

Texas A&M University - Department of Materials Science and Engineering

Langford Building A
798 Ross St.
College Station, TX 77843-3137
United States

Texas A&M University - Center for intelligent Multifunctional Materials and Structures

7607 Eastmark Dr
College Station, TX 77840
United States

Texas A&M University - Department of Mechanical Engineering

Langford Building A
798 Ross St.
College Station, TX 77843-3137
United States

Here is the Coronavirus
related research on SSRN

Paper statistics

Abstract Views
199
Downloads
18
PlumX Metrics