Application performs backpropagation algorithm for xor problem
X = [ 0, 0, 1, 1] Y = [ 0, 1, 0, 1] are inputs
Z = [ 0,1,1,0] is label array
n is learning rate
W's are weights and and initially gets a random value
W_new = W_old + ß
ß = n * (input) * (error)
global_error = label(Z) - output
hidden_error=sigmoid_derivative(input) * global_error * W_hidden * n