材料信息学是信息学技术在材料学中的应用,通过材料信息数据库和集成材料设计平台对材料的数据进行分析和预测。通过应用不同的机器学习(回归分析)方法和不同的特征选择算法,从众多的多尺度特征集中选择最优的特征子集可以预测金属氧化物的物理特性,归纳出适合材料不同特性的机器学习模型。分析结果表明,特征选择方法可以提升机器学习模型的性能,为进一步开发更有效的材料性能预测方法提供参考。
Material informatics is the application of information technology in material science,and the material data is analyzed and predicted by material information database and integrated material design platform.The main research was to predict the metal oxide physical properties by using different machine learning (regression analysis) methods and different feature selection algorithms to select the optimal feature subset from a large number of multi-scale feature sets.The most effective machine learning algorithm model was concluded for every material characteristic.At the same time,the performance improvement effect of feature selection algorithm was got for different machine learning models.It provided a reference for further development of more effective material performance prediction methods.
[1] 赵继成.材料基因组计划简介[J].自然杂志,2014,36(2):89-104.
[2] 王卓,王礞,雍歧龙,等.材料信息学及其在材料研究中的应用[J].中国材料进展,2017,36(2):132-140.
[3] 宋庆功,姜恩永,韩志勇,等.材料信息学平台的研究现状发展趋势与作用[J].材料导报,2008,22(1):70-73.
[4] Majzoub E H,Ozoliņš V.Prototype electrostatic ground state approach to predicting crystal structures of ionic compounds:application to hydrogen storage materials[J].Physical Review B,2008,77(10):104115.
[5] Wolverton C,Ozoliņš V,Asta M.Hydrogen in aluminum:first-principles calculations of structure and thermodynamics[J].Physical Review B,2016,69(14):1124-1133.
[6] Wolverton C,Ozoliņš V.Hydrogen storage in calcium alanate:first-principles thermodynamics and crystal structures[J].Physical Review B,2007,75(6):519-542.
[7] Meredig B,Agrawal A,Kirkin S,et al.Combinatorial screening for new materials in unconstrained composition space with machine learning[J].Physical Review B,2014,89(9):82-84.
[8] Takahashi K,Tanaka Y.Material synthesis and design from first principle calculations and machine learning[J].Computational Materials Science,2016,112:364-367.
[9] Fischer C C,Tibbetts K J,Morgan D,et al.Predicting crystal structure by merging data mining with quantum mechanics[J].Nature Materials,2006,5(8):641-646.
[10] Ceder G,Morgan D,Fischer C,et al.Data-mining-driven quantum mechanics for the prediction of structure[J].Mrs Bulletin,2006,31(12):981-985.
[11] Calfa B A,Kitchin J R.Property prediction of crystalline solids from composition and crystal structure[J].Aiche Journal,2016,62(8):2605-2613.
[12] Hart G L W.Evolutionary approach for determining first-principles model Hamiltonians[J].Nature Materials,2005,4(5):391-394.
[13] Jain A,Ong S P,Hautier G,et al.Commentary:the materials project:a materials genome approach to accelerating materials innovation[J].APL Materials,2013,1(1):1049.
[14] Ong S P,Cholia S,Jain A,et al.The materials application programming interface (API):a simple,flexible and efficient API for materials data based on representational state transfer (REST) principles[J].Computational Materials Science,2015,97(5):209-215.
[15] Ward L,Agrawal A,Choudhary A,et al.A general-purpose machine learning framework for predicting properties of inorganic materials[J].2016,2:16028.
[16] 王俊.基于原型方法的特征选择算法的研究[D].哈尔滨:哈尔滨工业大学,2009.
[17] Xiong Kai,Han Junwei.A review of regularization based feature selection algorithms[EB/OL].Beijing:Sciencepaper Online[2017-04-27].http://www.paper.edu.cn/releasepaper/content/201704-704.
[18] Tibshirani R.Regression shrinkage and selection via the lasso:a retrospective[J].Journal of the Royal Statistical Society,2011,73(3):273-282.
[19] Explained P M D.Introduction to linear regression analysis[J].Technometrics,2013,49(2):232-233.
[20] Breiman L.Random forests[J].Machine Learning,2001,45(1):5-32.
[21] 方匡南.随机森林组合预测理论及其在金融中的应用[M].厦门:厦门大学出版社,2012.
[22] 周志华.机器学习[M].北京:清华大学出版社,2016:178-181.
基金资助
国家自然科学基金项目(51741101)
PDF
(1524KB)
