Table 6 Comparison results with loss function, network depth, and autoencoder symmetry
Method | Model | RMSE ( ↓ ,0) | RSAM ( ↓ ,0) | MPSNR (↑,+∞) | MSSIM ( ↑ ,1) |
|---|---|---|---|---|---|
Loss Function | MoldSGR-AsyAutoencoder MAE | 0.02 | 0.02 | 35.32 | 0.81 |
MoldSGR-AsyAutoencoder Huber | 0.01 | 0.02 | 38.63 | 0.90 | |
aMoldSGR-AsyAutoencoder RMSE | 0.01 | 0.01 | 41.76 | 0.95 | |
Network Depth | Encoder 3/Decode 4 | 0.01 | 0.02 | 38.74 | 0.90 |
Encoder 4/Decode 3 | 0.02 | 0.03 | 36.88 | 0.85 | |
Encoder 3/Decode 3 | 0.02 | 0.03 | 36.87 | 0.85 | |
aMoldSGR-AsyAutoencoder | 0.01 | 0.01 | 41.76 | 0.95 | |
Autoencoder Symmetry | Symmetric Autoencoder (Encoder 4/Decode 4) | 0.01 | 0.02 | 38.63 | 0.90 |
Mirror Symmetric Autoencoder (Encoder 4/Decode 4) | 0.01 | 0.02 | 39.67 | 0.92 | |
aMoldSGR-AsyAutoencoder | 0.01 | 0.01 | 41.76 | 0.95 |
