Key features of neural networks
Hardware supports for Deep Neural Networks

Figure 2 shows the schematic diagram of a neural network. The network is broken down into layers: an input layer, one or more output layers and a number of intermediate layers, with no connection to the outside world, known as hidden layers. Each neuron in the hidden layers has a structure as shown in figure 3. A weighted combination of the different inputs (net_j) is compared with a threshold θ_j to activate or not the neuron's output. The most common activation functions are the sigmoid, the hyperbolic tangent and the ReLU function (figure 4)....