VLSI and Neural Networks

Abstract

This book mainly deals with theoretical aspects and the general analysis of neural networks. However, to study this new concept in detail and to exploit it for practical and perhaps beneficial purposes, one must face the challenge of performing real-life neural computations. Several courses can be followed to achieve this purpose. We enumerate the most important alternatives:

• 1. Simulation in software using sequential (von Neumann-type) computers.

• 2. Parallel computation using universal or dedicated special-purpose multiprocessor systems.

• 3. Realization of the network in hardware using digital or analog electronic components.

• 4. Realization of the network using nonlinear optical construction elements.