利用监督学习对含有非训练数据元素的化合物进行成分设计

IF 8.4 1区 材料科学 Q1 CHEMISTRY, PHYSICAL Journal of Materiomics Pub Date : 2024-07-14 DOI:10.1016/j.jmat.2024.06.008
Jingjin He , Ruowei Yin , Changxin Wang , Chuanbao Liu , Dezhen Xue , Yanjing Su , Lijie Qiao , Turab Lookman , Yang Bai
{"title":"利用监督学习对含有非训练数据元素的化合物进行成分设计","authors":"Jingjin He ,&nbsp;Ruowei Yin ,&nbsp;Changxin Wang ,&nbsp;Chuanbao Liu ,&nbsp;Dezhen Xue ,&nbsp;Yanjing Su ,&nbsp;Lijie Qiao ,&nbsp;Turab Lookman ,&nbsp;Yang Bai","doi":"10.1016/j.jmat.2024.06.008","DOIUrl":null,"url":null,"abstract":"<div><div>An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data. We show that the phase diagram of the ceramic (Ba<sub>1−<em>x</em>−<em>y</em></sub>Ca<sub><em>x</em></sub>Sr<sub><em>y</em></sub>)(Ti<sub>1−<em>u</em>−<em>v</em>−<em>w</em></sub>Zr<sub><em>u</em></sub>Sn<sub><em>v</em></sub>Hf<sub><em>w</em></sub>)O<sub>3</sub> can be accurately predicted if the feature values of unknown elements do not exceed the range of values for existing elements in the training data. In particular, we employ physical features as descriptors and compositions as weights to show that by excluding an element, such as Zr, Sn or Hf from the training set and treating it as an unknown element, the machine learning model accurately predicts the property only if the feature values of the unknown element does not exceed the range of values of existing elements in the training set. By adding a small amount of data for the unknown element restores the prediction accuracy. We demonstrate this for BaTiO<sub>3</sub> ceramics doped with rare earth elements where the prediction accuracy is restored if the physical feature space is suitably enlarged with training data. The prediction error increases with the Euclidean distance of the testing sample relative to the nearest training sample in the physical feature space. Our work provides an effective strategy for extending machine learning models for material compositions beyond the scope of available data.</div></div>","PeriodicalId":16173,"journal":{"name":"Journal of Materiomics","volume":"11 3","pages":"Article 100913"},"PeriodicalIF":8.4000,"publicationDate":"2024-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Compositional design of compounds with elements not in training data using supervised learning\",\"authors\":\"Jingjin He ,&nbsp;Ruowei Yin ,&nbsp;Changxin Wang ,&nbsp;Chuanbao Liu ,&nbsp;Dezhen Xue ,&nbsp;Yanjing Su ,&nbsp;Lijie Qiao ,&nbsp;Turab Lookman ,&nbsp;Yang Bai\",\"doi\":\"10.1016/j.jmat.2024.06.008\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data. We show that the phase diagram of the ceramic (Ba<sub>1−<em>x</em>−<em>y</em></sub>Ca<sub><em>x</em></sub>Sr<sub><em>y</em></sub>)(Ti<sub>1−<em>u</em>−<em>v</em>−<em>w</em></sub>Zr<sub><em>u</em></sub>Sn<sub><em>v</em></sub>Hf<sub><em>w</em></sub>)O<sub>3</sub> can be accurately predicted if the feature values of unknown elements do not exceed the range of values for existing elements in the training data. In particular, we employ physical features as descriptors and compositions as weights to show that by excluding an element, such as Zr, Sn or Hf from the training set and treating it as an unknown element, the machine learning model accurately predicts the property only if the feature values of the unknown element does not exceed the range of values of existing elements in the training set. By adding a small amount of data for the unknown element restores the prediction accuracy. We demonstrate this for BaTiO<sub>3</sub> ceramics doped with rare earth elements where the prediction accuracy is restored if the physical feature space is suitably enlarged with training data. The prediction error increases with the Euclidean distance of the testing sample relative to the nearest training sample in the physical feature space. Our work provides an effective strategy for extending machine learning models for material compositions beyond the scope of available data.</div></div>\",\"PeriodicalId\":16173,\"journal\":{\"name\":\"Journal of Materiomics\",\"volume\":\"11 3\",\"pages\":\"Article 100913\"},\"PeriodicalIF\":8.4000,\"publicationDate\":\"2024-07-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Materiomics\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2352847824001527\",\"RegionNum\":1,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"CHEMISTRY, PHYSICAL\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Materiomics","FirstCategoryId":"88","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2352847824001527","RegionNum":1,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CHEMISTRY, PHYSICAL","Score":null,"Total":0}
引用次数: 0

摘要图片

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Compositional design of compounds with elements not in training data using supervised learning
An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data. We show that the phase diagram of the ceramic (Ba1−xyCaxSry)(Ti1−uvwZruSnvHfw)O3 can be accurately predicted if the feature values of unknown elements do not exceed the range of values for existing elements in the training data. In particular, we employ physical features as descriptors and compositions as weights to show that by excluding an element, such as Zr, Sn or Hf from the training set and treating it as an unknown element, the machine learning model accurately predicts the property only if the feature values of the unknown element does not exceed the range of values of existing elements in the training set. By adding a small amount of data for the unknown element restores the prediction accuracy. We demonstrate this for BaTiO3 ceramics doped with rare earth elements where the prediction accuracy is restored if the physical feature space is suitably enlarged with training data. The prediction error increases with the Euclidean distance of the testing sample relative to the nearest training sample in the physical feature space. Our work provides an effective strategy for extending machine learning models for material compositions beyond the scope of available data.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Journal of Materiomics
Journal of Materiomics Materials Science-Metals and Alloys
CiteScore
14.30
自引率
6.40%
发文量
331
审稿时长
37 days
期刊介绍: The Journal of Materiomics is a peer-reviewed open-access journal that aims to serve as a forum for the continuous dissemination of research within the field of materials science. It particularly emphasizes systematic studies on the relationships between composition, processing, structure, property, and performance of advanced materials. The journal is supported by the Chinese Ceramic Society and is indexed in SCIE and Scopus. It is commonly referred to as J Materiomics.
期刊最新文献
Electronic state reconstruction enabling high thermoelectric performance in Ti doped Sb2Te3 flexible thin films Solar fuel photocatalysis Editor corrections to “Influence of electrode contact arrangements on polarisation-electric field measurements of ferroelectric ceramics: A case study of BaTiO3” [J Materiomics 11 (2025) 100939] Texture modulation of ferroelectric Hf0.5Zr0.5O2 thin films by engineering the polymorphism and texture of tungsten electrodes Graphical Contents list
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1