This repository represents the efforts in developing graph based multi-fidelity machine learning models for materials band gap prediction. We use the architecture of multi-fidelity graph networks[1] based on MEGNet[2]. We provide the models fitted with our efforts in the Models folder. We have published our data through a new environment, called the "Foundry", which will also allow model publication in the near future. The Foundry will (i) share models that are cloud accessible; (ii) host formatted data for easy access and use in ML; (iii) allow models and data to be updated with versioning. For more information about the Foundry, please visit https://github.com/MLMI2-CSSI/foundry#using-foundry-on-cloud-computing-resources. For the tutorials about how to extract the data from the Foundry and use the data to train the model, please visit example notebook in thie Notebook folder. The manuscipt associated with this work will coming soon.
- Chen, C.; Zuo, Y.; Ye, W.; Li, X.G.; Ong, S. P. Learning properties of ordered and disordered materials from multi-fidelity data. Nature Computational Science 2021, 1, 46–53 doi:10.1038/s43588-020-00002-x
- Chen, C.; Ye, W.; Zuo, Y.; Zheng, C.; Ong, S. P. Graph Networks as a Universal Machine Learning Framework for Molecules and Crystals. Chemistry of Materials 2019, 31(9), 3564-3572. doi:10.1021/acs.chemmater.9b01294