Microstructure classification in the unsupervised context
Microstructure classification in the unsupervised context
复制标题
无监督环境下的微观结构分类
DOI：
10.1016/j.actamat.2021.117434
复制
发表时间：
2022
期刊：
影响因子：
9.4
通讯作者：
Arróyave, Raymundo
中科院分区：
文献类型：
--
作者：
Kunselman, Courtney;Sheikh, Sofia;Mikkelsen, Madalyn;Attari, Vahid;Arróyave, Raymundo
传统的微结构分类需要由主题专家提供人工注释。人工输入的要求既昂贵又主观，并且不能跟上当前通过实验和计算产生的微结构图像的量。在这项工作中，我们开发了一个框架，能够通过利用新的机器学习过程来进行类发现和标签分配，从而降低这一过程中的人工标注成本。为了减少这种自动化过程造成的不良标签分配的惩罚，标签只被分配给高置信度的观测，而不明确的数据则被保留为无标签。然后使用半监督分类来利用高置信度和低置信度标签分配，并引入了一种新的将已建立的半监督误差估计技术推广到多类上下文中来评估所得到的分类器。最后，实验结果表明，该框架可以用来对仅通过数据驱动方法发现的微结构图像分类产生高精度的分类器，并且显示出类内一致的结构趋势和明显的类之间的形态差异。
Traditional microstructure classification requires human annotations provided by a subject matter expert. The requirement of human input is both costly and subjective and cannot keep up with the current volume of experimentally and computationally generated microstructure images. In this work, we develop a framework that is capable of reducing the cost of human annotation in this process by leveraging novel machine learning procedures for class discovery and label assignment. To reduce the penalty of a poor label assignment made by this automated process, labels are only assigned to high-confidence observations while ambiguous data are left unlabeled. Semi-supervised classification is then employed to leverage the high- and low-confidence label assignments, and a novel generalization of an established semi-supervised error estimation technique to the multi-class context is introduced to assess the resulting classifiers. Finally, it is shown that this framework can be used to produce highly accurate classifiers over microstructure image class taxonomies which are discovered solely through data-driven methods and which display consistent structural trends within and distinct morphological differences between classes.