¹⁷Any poles of H(s) that one would have associated with a neuron in the first of the two layers of the subnetwork can later easily be reintroduced without modifying the zeros of the subnetwork. This is done by copying the values of τ_1,ik and τ_2,ik of one of the two parallel neurons to the respective τ_1,ik and τ_2,ik of the other neuron. The two parallel neurons then have identical poles, which then also are the poles of any linearly weighted combination of their outputs. Poles associated with the neuron in the second of the two layers of the subnetwork are reintroduced without any special action.