Neural Network to Improve Crystal Structure Databases
A neural network has been trained to classify crystal structure errors in metal–organic frameworks (MOF) and other databases.
According to the study,
machine learning models are only as good as the data they are trained on, highlighting the importance of accurate databases.
The approach detects and classifies structural errors, including proton omissions, charge imbalances, and crystallographic disorder, to improve the fidelity of crystal structure databases.
This can help boost the accuracy of computational predictions used in materials discovery that rely on such databases.
Artificial intelligence and machine learning are becoming increasingly central to materials research, but concerns are growing over the reliability of the underlying datasets.
Author summary: Neural network improves crystal structure databases.