Graph Neural Networks Enhance Material Property Prediction for High-Entropy Oxides

Researchers have explored the use of graph neural networks (GNNs) for predicting the properties of high-entropy perovskite oxides (HEPOs), a complex class of materials. The study investigated ordered-to-disordered transfer learning, applying knowledge gained from chemically ordered perovskites to predict formation energy and the HOMO-LUMO gap in HEPOs. Results indicated that formation energy prediction transferred effectively, while HOMO-LUMO gap prediction showed limited transferability due to sensitivity to local chemical environments. Incorporating a small HEPO-specific dataset significantly improved HOMO-LUMO gap predictions, and the inclusion of three-body geometric information in GNN models enhanced the capture of complex structure-property relationships. AI

IMPACT Advances materials science research by improving predictive accuracy for complex material properties.

RANK_REASON Academic paper detailing a novel application of graph neural networks for materials science. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• Atomistic Line Graph Neural Network
• CGCNN
• GATGNN
• graph neural networks
• High-entropy perovskite oxides
• M3GNet
• Uniform Manifold Approximation and Projection

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →