Optical control of memristor dynamics for neuromorphic computing

点击翻译按钮获取中文摘要

In this project the student will build on recent work we have performed on silicon-based memristors in which we have shown how dynamic changes in memristor resistance can be used to perform computing tasks such as edge detection. Most recently, we have shown how optical stimulation can be used to control the dynamics of gradual resistance changes. The student will work to elucidate the mechanism(s) governing these optically-driven resistance changes and will look at how combined optical & electronic stimulation of memristors can be used to induce a range of oscillatory behaviours, which may be used to better mimic the behaviour of different classes of neuron, to generate random numbers, and to perform computational tasks. The project objectives are:- To characterise the degree of control of the dynamics of resistance changes that can be obtained using optical stimulation.- To understand the physical origin of the optical response.- To optimise material and device structure to maximise the combined optical and electrical response of memristors.- To illustrate, via an appropriate system-level demonstrator, how the response can be used to generate new functionality in memristor-based systems.- To develop a set of metrics to best characterise the novel systems (at both device and, importantly, system level), and to benchmark these against existing neuromorphic approaches.The aim will be to produce a prototype system demonstrator and to benchmark its performance against existing digital and analogue neuromorphic systems.