Convolutional encoders-decoders

Like other neural network (NN)-based systems, encoders-decoders benefited a lot from the introduction of convolutional and pooling layers. Deep auto-encoders (DAEs) and other architectures soon became widely used for increasingly complex tasks.

In this section, we will first introduce new layers developed for convolutional encoders-decoders. We will then present some significant architectures based on these operations.