Rough overview of VLSI systems, basic circuits, and technologies also offers some concrete examples of realized integrated circuits

Abstract

Artificial neural networks are systems based on special algorithms derived from the field of neuroscience. These networks display interesting features such as generalization, massive parallelism, learning, fault tolerance, and others. Until now, researchers have depended on computer simulations, sometimes conducted on dedicated computers, to generate most of their neural net research work. They used microprocessors to implement neural networks because of their small size and low price, high-level performance, and low power dissipation. Now, however, the authors realize that integrated neural networks are needed for decentralized or mobile systems and for robotics, prosthetics, or automotive applications in the expanding area of microelectronics. Neural networks, therefore, should also be available as microelectronic components.