Large-scale screening of 2D magnetic PMA candidates with ML incorporating crystal-field information

1/ AI use method: machine learning incorporating crystal-field information
2/ Applied task: classifying perpendicular magnetic anisotropy (PMA) in 2D magnetic materials and large-scale screening of candidates
3/ Reported result: achieved 89% classification accuracy and proposed 15 new PMA materials from more than 10,000 candidates
4/ This study classified perpendicular magnetic anisotropy (PMA) in 2D magnetic materials and screened large-scale candidates using machine learning that incorporates octahedral crystal-field information.
5/ According to the report, the classification accuracy was 89%, and by evaluating more than 10,000 candidates, 15 new PMA materials were proposed.
6/ In the materials field, such a classification model can be useful for quickly narrowing down computationally expensive searches for magnetic anisotropy.
7/ In particular, reflecting structural knowledge, here crystal-field information, in the features helps enable physically interpretable screening.
8/ However, a single accuracy value does not guarantee actual discovery yield, and the final validity of candidates must be confirmed through separate electronic-structure calculations or experiments.
9/ In addition, performance may deteriorate if the distribution of the screened candidate set differs from the training data. [1]