Table 5 Test MAE of different models for each of the materials properties for the prediction task of “Performance on Experimental Datasets”.

From: Improving deep learning model performance under parametric constraints for materials informatics applications

Dataset

Property

Size

AutoML

ElemNet

IRNet

BNet

BRNet

KEFE

Formation enthalpy (eV/atom)

2135

0.189

0.129

0.123

0.103

0.096

KEBG

Band gap (eV)

4604

0.464

0.468

0.493

0.483

0.460

  1. The lowest MAE values in each row are highlighted in bold.