Volatile memristors for neuromorphic hardware Dr. Pavel Borisov (Loughborough University)

Abstract
The recent rise of Chat GPT and other AI tools highlights even more the increased demand for more energy- and compute-efficient hardware. It is beneficial to start from a new, non von-Neumann concept and re-design hardware based on neuromorphic principles and by using new types of electronic devices which demonstrate functional similarity to biological neurons and synapses. Metal oxide memristors are one of those promising devices that are also compatible with the modern semiconducting technology. In this talk I will discuss how volatile memristors built from niobium oxide or silicon oxide films can be incorporated into artificial spiking neurons and how the corresponding spiking behaviour can be controlled by different parameters: temperature, voltage, radiation, strain. The second part of my talk will be devoted to discussion of how to apply niobium oxide memristor structures as physical reservoirs for reservoir computing tasks.

Fields
Thin film growth
Neuromorphic computing & Memristors
Magnetic, magnetoelectric and multiferroic materials

Online Lecture Access https://t1p.de/20240524_Lecture