Figure 1.

The semi‐supervised deep autoencoder structure. During the training, the deep autoencoder learns to reconstruct the input data and to predict the observed variables y, in this case, the subject's age and sex