The bias itself is like the coefficient of an extra input i.e., equal to 1. Since you are trying to calculate coefficient using neural networks.
output = 1.5*bias + 4*input1
So, the bias variable used here is like the coefficient (or weight) of an extra input i.e., 1.5. Now, if you consider 1.5 as an extra input, then you can obtain bias as the coefficient of the input = 1.5, in the same way of back propagation you used to obtain the coefficient of the input1 variable.
If you want to calculate “bias parameter” of the neural network using backpropagation, try using the form:
Output = weight*input + bias
