Analog implementations of artificial neural networks

Abstract

Analog implementations of artificial neural networks (ANN'S) will be discussed. Analog ANN'S should be faster and smaller than digital implementations, however, the problem of the smaller dynamic range of analog storage must be addressed. In addition, long-term memory is not as easy to implement in analog circuits as it is in digital. The prospects of several different techniques for implementing analog ANN'S are presented along with a brief survey of recent research results.

Tuneable Current Mode RMS Detector

“…In many applications, particularly in sensor signal processing, the inputs and outputs need to be analogue so that the unlimited resolution and accuracy that is possible with purely digital circuitry will not be available [40].…”

A New Low Power Four Quadrant Analog Multiplier

“…Signals can be currents or voltages due to the demand [1]. As analog multiplication is important in many operations so it has a wide of applications in many fields like adaptive filters, equalizer, modulators, automatic gain controlling, artificial neural networks [2,3,4], image processing and medicine [5]. These days, the electronic society is faced with orders in relation to low power dissipations [4,6].…”