In the previous chapters, we covered feedforward NNs, CNNs, and RNNs. These networks are predominantly used for supervised learning tasks. In this chapter, we focus on autoencoders (see Figure 11-1), a neural network architecture which is mainly used for unsupervised learning tasks.
The main promise of the autoencoders is to learn an encoding structure and a decoding structure for a given set of data. Autoencoders are mainly used for dimensionality reduction, noise reduction, and several generative tasks. There are several variants of autoencoders designated for particular ...
