Deep neural networks
We could define DL as a class of ML techniques, where information is processed in hierarchical layers to understand representations and features from data in increasing levels of complexity. In practice, all DL algorithms are NNs, which share some common basic properties. They all consist of a graph of interconnected operations, which operate with input/output tensors. Where they differ is network architecture (or the way units are organized in the network), and sometimes in the way they are trained. With that in mind, let’s look at the main classes of NNs. The following list is not exhaustive, but it represents most NN types in use today:
- Multilayer perceptron (MLP): An NN with feedforward propagation, fully connected layers, and at least one hidden layer. We introduced MLPs in Chapter 2.
- Convolutional neural network (CNN): A CNN is a feedforward NN with several types of special layers. For example, convolutional layers apply a filter to the...
