Well then, what's so different about the idea of autoencoders? You have surely come across countless encoding algorithms, ranging from MP3 compression that's performed to store audio files, or JPEG compression to store image files. The reason autoencoding neural networks are interesting is they take a very different approach toward representing information compared to their previously stated quasi-counterparts. It is the kind of approach you have certainly come to expect after seven long chapters on the inner workings of neural networks.

Unlike the MP3 or JPEG algorithms, which hold general assumptions about sound and pixels, a neural autoencoder is forced to learn representative features automatically from whatever input it is shown during a training session. It proceeds to recreate the given input by using the learned representations...