E N D
In step 1, a neural network is set up. In addition to setting up the structure or topology of the network, random numbers between 1 and 1 are assigned as weights to each of the connections. For example, the weight from node A to node D (wAD) is 0.4 and the weight from node E to F (wEF) is 0.1
In step 2, a random observation is selected (v) and is presented to the network as shown. The value of I1 (0) is presented to A, the value of I2 (1) is presented to B, and the value of I3 (1) is presented to C.
In step 3, these inputs in combination with the connection weights are used to calculate the output from the hidden nodes D and E. To calculate these outputs, nodes D and E first combine the inputs and then use an activation function to derive the outputs. The combined inputs to nodes D and E are the weighted sum of the input values
In step 4, the outputs from D and E are used as inputs to F. The total input is calculated by combining these values with the weights of the connections: