Fig. 4From: Deep convolutional architectures for extrapolative forecasts in time-dependent flow problemsMLP autoencoder architectureBack to article page