The problems arising from the use of nonlinear multipliers in multilay
er neural network synapse structures are discussed. The errors arising
from the neglect of nonlinearities are shown and the effect of traini
ng in eliminating these errors is discussed. A method for predicting t
he final errors resulting from nonlinearities is described. Our approx
imate results are compared with the results from circuit simulations o
f an actual multiplier circuit.