Machine Learning versus Human Learning in Predicting Glass-Forming Ability of Metallic Glasses
Machine Learning versus Human Learning in Predicting Glass-Forming Ability of Metallic Glasses
复制标题
DOI：
10.1016/j.actamat.2022.118497
复制
发表时间：
2022-11
期刊：
影响因子：
9.4
通讯作者：
Guannan Liu;S. Sohn;Sebastian A. Kube;Arindam Raj;Andrew Mertz;A. Nawano;Anna Gilbert;M. Shattuck;C. O’Hern;J. Schroers
中科院分区：
文献类型：
--
作者：
Guannan Liu;S. Sohn;Sebastian A. Kube;Arindam Raj;Andrew Mertz;A. Nawano;Anna Gilbert;M. Shattuck;C. O’Hern;J. Schroers
复杂的材料科学问题，如玻璃的形成，必须考虑大的系统尺寸，这是许多数量级太大，无法解决的第一原理计算。机器学习（ML）在其他各个领域的成功应用表明，ML可能有助于解决材料科学中的复杂问题。为了测试其有效性，我们试图使用ML预测块体金属玻璃的形成。令人惊讶的是，我们发现，最近开发的ML模型的基础上，201合金的功能，使用31个元素的功能的简单组合构造的模型是无法区分的非物理特征的基础上。只有当训练和测试数据实现显著分离时，201ML模型才比非物理模型表现得更好。然而，它的表现明显不如基于人类学习的三特征模型。201ML模型的有限性能源于无法通过元素特征准确表示合金特征，这意味着需要对混合行为的物理见解来开发可预测的ML模型。
Complex materials science problems such as glass formation must consider large system sizes that are many orders of magnitude too large to be solved by first-principles calculations. The successful application of machine learning (ML) in various other fields suggests that ML could be useful to address complex problems in materials science. To test its efficacy, we attempt to predict bulk metallic glass formation using ML. Surprisingly, we find that a recently developed ML model based on 201 alloy features constructed using simple combinations of 31 elemental features is indistinguishable from models that are based on unphysical features. The 201ML-model performs better than the unphysical models only when significant separation of training and testing data is achieved. However, it performs significantly worse than a human-learning based three-feature model. The limited performance of the 201ML-model originates from the inability to accurately represent alloy features through elemental features, which means that physical insights about mixing behavior are required to develop predictable ML-models.