Fig. 5From: Deep convolutional architectures for extrapolative forecasts in time-dependent flow problemsConvolutional autoencoder architectureBack to article page