Review of memristor based neuromorphic computation: opportunities, challenges and applications

Abstract

The memristor is regarded as one of the promising possibilities for next-generation
computing systems due to its small size, easy construction, and low power consumption.
Memristor-based novel computing architectures have demonstrated
considerable promise for replacing or enhancing traditional computing platforms
that encounter difficulties in the big-data era. Additionally, the striking resemblance
between the mechanisms governing the programming of memristance and
the manipulation of synaptic weight at biological synapses may be used to create
unique neuromorphic circuits that function according to biological principles.
Nevertheless, getting memristor-based computing into practice presents
many technological challenges. This paper reviews the potential for memristor
research at the device, circuit, and system levels, mainly using memristors to
demonstrate neuromorphic computation. Here, the common issues obstructing
the development and widespread use of memristor-based computing systems are
also carefully investigated. This study speculates on the prospective applications
of memristors, which can potentially transform the field of electronics altogether.