Spintronic materials and devices towards an artificial neural network: accomplishments and the last mile

Abstract

ABSTRACTNeuromorphic computing based on brain-inspired frameworks has shown its potential to perform computation with remarkable power and speed efficiency. The spintronic device has been considered a promising candidate for neuromorphic computing because of its intrinsic non-volatility, ultrafast switching dynamics and scalability. This article will review the current status of neuromorphic computing techniques employing spintronic materials and devices. We will begin with a concise introduction to the biological structure of neurons and then go through several important models of artificial neural networks. The revolutionary breakthroughs of spintronic-based neuromorphic computation will be demonstrated together with their challenges and outlook.

Keywords

• spintronics
• magneto-resistive random-access memory
• neuromorphic computing
• artificial neural network