中国物理B ›› 2021, Vol. 30 ›› Issue (5): 50705-050705.doi: 10.1088/1674-1056/abf12d

所属专题: SPECIAL TOPIC — Machine learning in condensed matter physics

• • 上一篇    下一篇

Efficient sampling for decision making in materials discovery

Yuan Tian(田原)1, Turab Lookman2,†, and Dezhen Xue(薛德祯)1,‡   

  1. 1 State Key Laboratory for Mechanical Behavior of Materials, Xi'an Jiaotong University, Xi'an 710049, China;
    2 Los Alamos National Laboratory, Los Alamos, NM 87545, USA
  • 收稿日期:2020-12-08 修回日期:2021-01-17 接受日期:2021-03-24 出版日期:2021-05-14 发布日期:2021-05-14
  • 通讯作者: Turab Lookman, Dezhen Xue E-mail:turablookman@gmail.com;xuedezhen@xjtu.edu.cn
  • 基金资助:
    Project supported by the National Key Research and Development Program of China (Grant No. 2017YFB0702401) and the National Natural Science Foundation of China (Grant Nos. 51571156, 51671157, 51621063, and 51931004).

Efficient sampling for decision making in materials discovery

Yuan Tian(田原)1, Turab Lookman2,†, and Dezhen Xue(薛德祯)1,‡   

  1. 1 State Key Laboratory for Mechanical Behavior of Materials, Xi'an Jiaotong University, Xi'an 710049, China;
    2 Los Alamos National Laboratory, Los Alamos, NM 87545, USA
  • Received:2020-12-08 Revised:2021-01-17 Accepted:2021-03-24 Online:2021-05-14 Published:2021-05-14
  • Contact: Turab Lookman, Dezhen Xue E-mail:turablookman@gmail.com;xuedezhen@xjtu.edu.cn
  • Supported by:
    Project supported by the National Key Research and Development Program of China (Grant No. 2017YFB0702401) and the National Natural Science Foundation of China (Grant Nos. 51571156, 51671157, 51621063, and 51931004).

摘要: Accelerating materials discovery crucially relies on strategies that efficiently sample the search space to label a pool of unlabeled data. This is important if the available labeled data sets are relatively small compared to the unlabeled data pool. Active learning with efficient sampling methods provides the means to guide the decision making to minimize the number of experiments or iterations required to find targeted properties. We review here different sampling strategies and show how they are utilized within an active learning loop in materials science.

关键词: sampling methods, active learning, decision making, material design, Bayesian optimization

Abstract: Accelerating materials discovery crucially relies on strategies that efficiently sample the search space to label a pool of unlabeled data. This is important if the available labeled data sets are relatively small compared to the unlabeled data pool. Active learning with efficient sampling methods provides the means to guide the decision making to minimize the number of experiments or iterations required to find targeted properties. We review here different sampling strategies and show how they are utilized within an active learning loop in materials science.

Key words: sampling methods, active learning, decision making, material design, Bayesian optimization

中图分类号:  (Design of experiments)

  • 07.05.Fb
07.05.Mh (Neural networks, fuzzy logic, artificial intelligence) 81.90.+c (Other topics in materials science) 07.05.Tp (Computer modeling and simulation)