CrysXPP:An explainable property predictor for crystalline materials

作     者：Kishalay Das Bidisha Samanta Pawan Goyal Seung-Cheol Lee Satadeep Bhattacharjee Niloy Ganguly 

作者机构：Indian Institute of Technology KharagpurKharagpurIndia Indo Korea Science and Technology CenterBangaloreIndia Leibniz University of HannoverHannoverGermany 

出 版 物：《npj Computational Materials》 (计算材料学（英文）)

年 卷 期：2022年第8卷第1期

页      面：424-434页

核心收录：

学科分类：07[理学] 0701[理学-数学] 070101[理学-基础数学] 

基　　金：Projekt DEAL 

主　　题：materials. property crystalline 

摘      要：We present a deep-learning framework,CrysXPP,to allow rapid and accurate prediction of electronic,magnetic,and elastic properties of a wide range of *** lowers the need for large property tagged datasets by intelligently designing an autoencoder,*** important structural and chemical properties captured by CrysAE from a large amount of available crystal graphs data helped in achieving low prediction ***,we design a feature selector that helps to interpret the model’s *** notably,when given a small amount of experimental data,CrysXPP is consistently able to outperform conventional DFT.A detailed ablation study establishes the importance of different design *** release the large pre-trained model *** believe by fine-tuning the model with a small amount of property-tagged data,researchers can achieve superior performance on various applications with a restricted data source.