Architecture of Artificial neural networks:

Presumably "Artificial Neural Networks" consist of a number of units that are mini calculation devices. But they take in real-valued input from multiple other nodes and they produce a single real valued output.  Through real-valued input and output we mean real numbers that are able to take any decimal value. However the architecture of ANNs is as given: 

1. There a set of input units that take in information about the example to be propagated by the network. Through propagation the information from the input will be passed by the network and an output produced. Presumably the set of input units forms that what is known as the input layer. 

2. There a set of hidden units that take input from the input layer. If there the hidden units collectively form the  hidden layer then just to simplicity here we assume like each unit in the input layer is connected to each unit of the hidden layer so it means this isn't necessarily the case. As a weighted su units forms the input to every hidden unit. Remember that the number of hidden units is usually smaller than the number of input units of course. 

3. There a set of output units that in learning tasks and dictate the category assigned to an example propagated through the network. So the output units form the output layer. Hence again for simplicity here we assume that each unit in the hidden layer is connected to each unit in the output layer so a weighted sum of the output from the hidden units forms the input to every output unit.  Still m of the output from the input.