Refer to the following multilayer perceptron network with two hidden layers. All nodes use a step activation function, i$.$e$.,$

output $=+1$ if total input $>=$ bias

output $=-1$ if total input $<$ bias