Neural Networks with R

The network training phase

The artificial neural networks are composed of simple elements operating in parallel. Connections between network elements are fundamental as they decide network functions. These connections affect the result through its weight, which is regulated in the neural network training phase. In the following diagram is shown a comparison between serial and parallel processing:

Then, in the training phase, the network is regulated by changing the connection weights, so that a particular input will lead to a specific destination. For example, the network can be adjusted by comparing the output (what we calculate practically) ...