April 2018
Intermediate to advanced
334 pages
10h 18m
English
A deep autoencoder is a type of deep neural network composed of two symmetrical neural networks, as shown in the following diagram, which is capable of converting input data to a more compact representation that is also lower in dimension. The encoder network first encodes the input into a compact compressed representation and the decoder network decodes that representation back to output the original input. As shown in the following diagram, there are two neural networks (encoder and decoder) connected by a middle layer, which contains the compact compressed representation of the input data:
