About this document
Local Augmentation for GNNs in Materials by manoranjan sahu is a document available to read on EtoBox.
This article discusses a method to enhance the predictive accuracy of graph neural networks (GNNs) for unrelaxed structures in computational materials discovery by employing a data-augmentation technique. By perturbing the atomic coordinates of relaxed structures, the authors demonstrate a significant reduction in prediction error for formation energies, improving accuracy by 66%. The proposed approach not only enhances the model
- Author
- manoranjan sahu
- Language
- EN