Suppose you have a neural network with 3 layers. The first layer is the input layer with 20 neurons just encoding the 20 input values (taking values between 0 and 1). The second layer consists of 30 sigmoid neurons each with their own weights and biases. The final layer is just one output sigmoid neuron with its weights and bias. The neurons in adjacent layers are fully connected. So this neural network computes a function y=f(x₁,x₂,,x₂₀) where all variables take values between 0 and 1. Is it possible to change the weights and biases of the sigmoid neurons in this neural network in a systematic way so that the new neural network computes a new function z=h(x₁,x₂,,x₂₀) such that h(x₁,x₂,,x₂₀)=1- f(x₁,x₂,,x₂₀) for all possible inputs x₁,x₂,,x₂₀? If so, how?