Robust Memristor Networks for Neuromorphic Computation Applications.

Dániel Hajtó,György Cserey,Ádám Rák

doi:10.3390/ma12213573

Dániel Hajtó, György Cserey + Show 1 more

Open Access

https://doi.org/10.3390/ma12213573

Copy DOI

Abstract

One of the main obstacles for memristors to become commonly used in electrical engineering and in the field of artificial intelligence is the unreliability of physical implementations. A non-uniform range of resistance, low mass-production yield and high fault probability during operation are disadvantages of the current memristor technologies. In this article, the authors offer a solution for these problems with a circuit design, which consists of many memristors with a high operational variance that can form a more robust single memristor. The proposition is confirmed by physical device measurements, by gaining similar results as in previous simulations. These results can lead to more stable devices, which are a necessity for neuromorphic computation, artificial intelligence and neural network applications.

Highlights

Since the theoretical [1] and practical [2] discovery of memristors, they have been extensively studied [3,4,5] as elementary building blocks for artificial intelligence and neuromorphic computing applications
Research has been done [6] to find optimal materials that satisfy these expectations, but even there are other possibilities to further increase the capabilities of memristors
The measured memristor devices are made of Ge2 Se3 and Ag based chalcogenide dielectric with W (Tungsten) conductors

Summary

Introduction

Since the theoretical [1] and practical [2] discovery of memristors, they have been extensively studied [3,4,5] as elementary building blocks for artificial intelligence and neuromorphic computing applications. The expected properties of memristors for such applications are wide and analog resistance range, low variance of device parameters and high device stability during long-term operation. Research has been done [6] to find optimal materials that satisfy these expectations, but even there are other possibilities to further increase the capabilities of memristors. The first one is having two clearly distinguishable states and these state declarations should apply to every element in a memory array. To reach the performance of the current complementary metal–oxide–semiconductor (CMOS) technology’s

Objectives

Methods

Discussion

Conclusion