Artificial Neural Networks
Figure 1: Neuron of the human brain

The artificial neural network tries to simulate the activities of the neurons like that of the human brain.

An artificial neuron is the most basic component of the network. Neurons are organized in such a way that facilitates the flow and the processing of data (the multi layer perceptron) from layer to layer. Through this process, the data are evaluated simultaneously and weighted to find the best fit that would reflect an accurate prediction, as it relates to the training data. These weights (wn) may differ for each input signal (xn) of an activation function (O), for the calculation of the output signal. Activation functions are often defined by the sigmoid curve.

IO Neuron Figure 2: Input flow of an artificial neuron  

The output signal of each neuron of a previous layer is the input signal for each neuron of the next layer. All input signals for one neuron are weighted and used for the calculation of the output signal. The number of layers and neurons per layer can differ for each scenario. This type of network is called Multi Layer Perceptron .

By an iterative process, the software tries to find the minimum error between the expected and the calculated results.

  Sigmoid function Figure 3: Sigmoid function