Implementation of FNNS using simple nonlinear circuits

Abstract

Simple nonlinear circuits are proposed for implementing feedforward neural networks with learning. A simple nonlinear multiplier circuit and a simple nonlinear difference circuit have been designed. FNN circuits consist of multi-layered feed forward circuits and learning circuitry, which are implemented by using nonlinear synapse circuits, sigmoid circuits, and nonlinear multipliers. The learning circuitry is implemented by employing MEBP (Modified Error Back-Propagation) learning rule. The proposed FNNs produce an output voltage, which is uniquely determined by any pair of learning input pattern. The proposed FNNs are applied for two-layer feedforward neural network model and their operations have been verified by using HSPICE circuit simulator The proposed FNNs are very suitable for the future implementation of the large-scale neural networks with learning.