An Electronically Implemented Neural Network for Real-Life Applications

A hybrid neural network implemented in VLSI hardware for real-life applications is presented. This is achieved by using the simple circuits in electronic industry. This network contains both RBF and MLP structures in one single network and it can be used in many application areas, such as pattern recognition, character recognition, classification problems, signal processing, medical diagnosis, and so on. As an application, a simple classification problem was considered and a demonstration circuit has been designed to solve this problem. The circuit designed has been simulated using cdsSpice simulator in the Cadence design package on a Sun workstation to show the functionality of the network.