Table 1 Performance of GNN models on the BDG dataset

From: Adaptive edge-aware graph convolutional with multi-task learning for simultaneous prediction of material properties

Model

MAE↓ Egap

RMSE↓ Egap

R2 ↑ Egap

MAE↓ WF

RMSE↓ WF

R2 ↑ WF

GATConv

0.0076

0.0172

0.9898

0.0155

0.0355

0.9683

GraphSAGE

0.0127

0.0237

0.9819

0.0299

0.0506

0.9354

GINConv

0.0123

0.0249

0.9800

0.0185

0.0324

0.9736

GCNII

0.0349

0.0459

0.9343

0.0501

0.0668

0.8939

SchNet

0.0097

0.0271

0.9640

0.0788

0.1034

0.7472

AEGCNN-MTL

0.0031

0.0131

0.9905

0.0073

0.02700

0.9778

  1. Reported metrics are mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R²) for band gap (Egap) and work function (WF). The downward arrow indicates that a lower value denotes better performance, while the upward arrow indicates that a higher value denotes better performance. The best results for each metric are highlighted in bold.
Back to article page