Machine learning(ML)is widely used to explore crystal materials and predict their properties.However,the training is time-consuming for deep-learning models,and the regression process is a black box that is hard to in...Machine learning(ML)is widely used to explore crystal materials and predict their properties.However,the training is time-consuming for deep-learning models,and the regression process is a black box that is hard to interpret.Also,the preprocess to transfer a crystal structure into the input of ML,called descriptor,needs to be designed carefully.To efficiently predict important properties of materials,we propose an approach based on ensemble learning consisting of regression trees to predict formation energy and elastic constants based on small-size datasets of carbon allotropes as an example.Without using any descriptor,the inputs are the properties calculated by molecular dynamics with nine different classical interatomic potentials.Overall,the results from ensemble learning are more accurate than those from classical interatomic potentials,and ensemble learning can capture the relatively accurate properties from the nine classical potentials as criteria for predicting the final properties.展开更多
基金Funding for this research was provided by National Science Foundation (NSF) under award numbers CMMI-1826439, CMMI-1762792 and CMMI-1825739. This support is greatly acknowledged. The authors also thank the Agave Computer Cluster of ASU for providing the computational resources.
文摘Machine learning(ML)is widely used to explore crystal materials and predict their properties.However,the training is time-consuming for deep-learning models,and the regression process is a black box that is hard to interpret.Also,the preprocess to transfer a crystal structure into the input of ML,called descriptor,needs to be designed carefully.To efficiently predict important properties of materials,we propose an approach based on ensemble learning consisting of regression trees to predict formation energy and elastic constants based on small-size datasets of carbon allotropes as an example.Without using any descriptor,the inputs are the properties calculated by molecular dynamics with nine different classical interatomic potentials.Overall,the results from ensemble learning are more accurate than those from classical interatomic potentials,and ensemble learning can capture the relatively accurate properties from the nine classical potentials as criteria for predicting the final properties.
