Neural network control over operation accuracy of memristor-based hardware

Abstract

A general approach to controlling the operation accuracy of memristor-based hardware (MBH) is proposed herein. The following approach is based on application of the neural network algorithms, which make it possible to register the excess of the allowed inaccuracy level of signal processing in MBH. The artificial neural networks of radial basis functions meant to control the level of additive noises of pulse frequency modulated signals in MBH have been designed and studied. The results of practical application of the deigned algorithms are shown herein.