Let's recap on what we learned in Chapter 4, Training Neural Networks. A neural network is a machine emulation of the human brain that is seen as a set of algorithms that have been set out to extract patterns out of data. It has got three different layers:

• Input layer
• Hidden layer
• Output layer

Sensory numerical data (in the form of a vector) passes through the input layer and then goes through the hidden layers to generate its own set of perceptions and reasoning to yield the final result in the output layer.

Can you recall what we learned in Chapter 4, Training Neural Networks, regarding the number of layers in ANN and how we count them? When we have got the layers like the ones shown in the following diagram, can you count the number of layers? Remember, we always count just the hidden layer and the output layer. So, if somebody is asking you how many layers there are in your network, you don't include the input layer while answering:

Yes, that's right—there are two layers in the preceding architecture. What about for the following network?

This network has got three layers, which includes two hidden layers. As the layers increase, the model becomes deeper.