Skip to main content
. 2016 Jan 19;11(1):e0146672. doi: 10.1371/journal.pone.0146672

Fig 3.

Fig 3

(a) shows the architecture of the autoencoder, whose input and output layers have the same size despite having smaller hidden layers. The bottom network performs the role of an encoder, whereas the top network performs the role of a decoder; (b) geometrical representation of the LEAE.