Compositionally-Restricted Attention-Based Network (CrabNet)

This software package implements the Compositionally-Restricted Attention-Based Network (CrabNet) that takes only composition information to predict material properties. Additionally, it demonstrates several model interpretability techniques that are possible with CrabNet.

Table of Contents

• Publications / How to cite
• Installation and basic use of CrabNet
• Model interpretability with CrabNet
• Maintainers

Publications / How to cite

This repository contains the code accompanying two CrabNet publications. Please consider citing them if you want to use CrabNet or the techniques discussed within the works:

1. The "CrabNet publication": A. Y.-T. Wang, S. K. Kauwe, R. J. Murdock, T. D. Sparks, Compositionally restricted attention-based network for materials property predictions, npj Comput. Mater., 2021, 7: 77. DOI: 10.1038/s41524-021-00545-1.
2. The "ExplainableGap publication": A. Y.-T. Wang, M. S. Mahmoud, M. Czasny, A. Gurlo, CrabNet for Explainable Deep Learning in Materials Science: Bridging the Gap Between Academia and Industry, Integr. Mater. Manuf. Innov., 2022, 11 (1): 41-56. DOI: 10.1007/s40192-021-00247-y.

The references in bibtex form:

@article{Wang2021crabnet,
 author = {Wang, Anthony Yu-Tung and Kauwe, Steven K. and Murdock, Ryan J. and Sparks, Taylor D.},
 year = {2021},
 title = {Compositionally restricted attention-based network for materials property predictions},
 pages = {77},
 volume = {7},
 number = {1},
 doi = {10.1038/s41524-021-00545-1},
 publisher = {{Nature Publishing Group}},
 shortjournal = {npj Comput. Mater.},
 journal = {npj Computational Materials}
}

@article{Wang2022explainablegap,
 author = {Wang, Anthony Yu-Tung and Mahmoud, Mahamad Salah and Czasny, Mathias and Gurlo, Aleksander},
 year = {2022},
 title = {CrabNet for Explainable Deep Learning in Materials Science: Bridging the Gap Between Academia and Industry},
 url = {https://doi.org/10.1007/s40192-021-00247-y},
 pages = {41--56},
 volume = {11},
 number = {1},
 shortjournal = {Integr. Mater. Manuf. Innov.},
 journal = {Integrating Materials and Manufacturing Innovation},
 doi = {10.1007/s40192-021-00247-y},
 publisher = {{Springer International Publishing AG}}
}

Installation and basic use of CrabNet

For the steps, please see the readme located at README_CrabNet.md.

Model interpretability with CrabNet

For the steps, please see the readme located at README_ExplainableGap.md.

Maintainers

• sgbaird, Sterling Baird (main maintainer)
• anthony-wang, Anthony Wang