From: Deep convolutional architectures for extrapolative forecasts in time-dependent flow problems
Hyperparameters | MLP AE | Convolutional CAE |
---|---|---|
Encoder layers | [500, 250] | [8, 32, 32] |
Latent dimension (m) | 25, 50, 125 | 25,50,125 |
Activation | relu, swish | relu, swish |
Loss function | MSE | MSE |
Learning rate | \(10^{-3},3 \times 10^{-4}\) | \(10^{-3},3 \times 10^{-4}\) |
Model parameters (\(m = 125\)) | 1,265,025 | 13,634 |