There is growing interest in applying machine learning techniques in the field of materials science.How-ever,the interpretation and knowledge extracted from machine learning models is a major concern,par-ticularly as formulating an explicit model that provides insight into physics is the goal of learning.In the present study,we propose a framework that utilizes the filtering ability of feature engineering,in con-junction with symbolic regression to extract explicit,quantitative expressions for the band gap energy from materials data.We propose enhancements to genetic programming with dimensional consistency and artificial constraints to improve the search efficiency of symbolic regression.We show how two de-scriptors attributed to volumetric and electronic factors,from 32 possible candidates,explicitly express the band gap energy of NaCl-type compounds.Our approach provides a basis to capture underlying phys-ical relationships between materials descriptors and target properties.

Beijing Advanced Innovation Center for Materials Genome Engineering,University of Science and Technology Beijing,Beijing 100083,China;Corrosion and Protection Center,University of Science and Technology Beijing,Beijing 100083,China;Research Center for Materials Genome Engineering,Sichuan University,Chengdu 610065,China;AiMaterials Research LLC,Santa Fe,NM 87501,United States of America;School of Materials Science and Engineering,Harbin Institute of Technology,Shenzhen,China;Materials Genome Institute,Shanghai University,333 Nanchen Road,Shanghai 200444,China

[1]Changxin Wang;Yan Zhang;Cheng Wen;Mingli Yang;Turab Lookman;Yanjing Su;Tong-Yi Zhang-.Symbolic regression in materials science via dimension-synchronous-computation)[J].材料科学技术（英文版）,2022(27):77-83

scriptors

Symbolic,regression,materials,science,via,synchronous,computation,There,growing,interest,applying,machine,learning,techniques,field,How,ever,interpretation,knowledge,extracted,from,models,major,concern,par,ticularly,formulating,that,provides,insight,into,physics,goal,In,present,study,propose,framework,utilizes,filtering,ability,feature,engineering,junction,symbolic,quantitative,expressions,band,gap,energy,data,We,enhancements,genetic,programming,dimensional,consistency,artificial,constraints,improve,search,efficiency,show,two,attributed,volumetric,electronic,factors,possible,candidates,explicitly,NaCl,type,compounds,Our,approach,basis,capture,underlying,ical,relationships,between,descriptors,target,properties

0.639941