The first step of this update function is to search for the first hidden unit of the network. The current output is saved and a new output is calculated for all neurons of the hidden and output layer. Once this is accomplished the next progression of the hidden and output units starts. Now for each neuron of the hidden and output layer the new output is saved and the old saved output is restored. With this older output the activation of all hidden and output neurons is calculated. After this task is accomplished the new saved output value of all hidden and output neurons is restored.