Table 1 Validation loss and approximate time to train the model for different benchmarked models.
From: A scalable convolutional neural network approach to fluid flow prediction in complex environments
Architecture | Parameters | MSE | Time (min) |
|---|---|---|---|
U-net | 1,081,914 | 0.0471 | 90 |
Residual U-net | 1,475,946 | 0.0350 | 200 |
Gated residual U-net | 2,016,858 | 0.0324 | 240 |
Nested U-net | 566,586 | 4.2372 | 330 |
Recurrent Residual U-net | 3,306,154 | 0.0305 | 270 |
