Table 2 Performance comparison between the CNN models in terms of the composition of the loss function.

From: Simulation acceleration for transmittance of electromagnetic waves in 2D slit arrays using deep learning

Model

Loss function

RMSE

R2 score

Train

Test

Train

Test

CNN

RMSE

0.0364

0.0629

0.9496

0.8551

RMSE + diff. RMSE

0.0349

0.0633

0.9542

0.8534

RMSE + diff. BCE

0.0370

0.0619

0.9480

0.8601

MLP

RMSE

0.0630

0.0837

0.8485

0.7438

Random forest

RMSE

0.0416

0.0807

0.9372

0.7605

Extra tree

RMSE

0.0292

0.0880

0.9693

0.7178

  1. Root mean squared errors and R2 scores were measured from the overall output values of the train set and test set.
Back to article page