Table 12 2D River problem: hyperparameters for the 2D CAE network
From: Deep convolutional architectures for extrapolative forecasts in time-dependent flow problems
Hyperparameters | 2D convolutional CAE |
|---|---|
Encoder layers channels | [16, 32, 16] |
padding mode | Replicate |
Latent dimension (m) | 32x32 |
Activation | relu |
Loss function | MSE |
Learning rate | \(3 \times 10^{-4}\) |
Model parameters (\(m = 32 \times 32\)) | 19,138 |