A MULTILEVEL NEURAL-NETWORK FOR A D CONVERSION

In this paper we introduce a multilevel neuron and show its use in a n eural network multilevel A/D converter. An energy function suited for multilevel neural networks is defined for which local minima problems for A/D conversion are removed by modifying Lee and Sheu's method. Thi s energy function extends others in the sense that it allows one to co nsider more than two discrete levels in the neuron output and threshol d settings. We also demonstrate how to build and implement multilevel nonlinearities, and a way of implementing a multilevel neural network for A/D conversion by taking advantage of BiCMOS technologies. Compute r simulations are included to illustrate how this design functions and individual component VLSI chips measurements for multilevel A/D conve rsion are presented to show how each component operates.