Transfer learning on large datasets for the accurate prediction of material properties
Noah Hoffmann, Jonathan Schmidt, Silvana Botti, Miguel A. L. Marques
Abstract
Pretraining on large, lower-fidelity datasets enables extremely effective training of graph neural networks on smaller, high-fidelity datasets.
Topics & Concepts
FidelityTransfer of learningComputer scienceArtificial intelligenceGraphHigh fidelityMachine learningArtificial neural networkTraining setTheoretical computer scienceEngineeringTelecommunicationsElectrical engineeringMachine Learning in Materials ScienceComputational Drug Discovery MethodsFuel Cells and Related Materials