Radio-Frequency Spintronic Neural Networks

• Published in: 2023 IEEE International Magnetic Conference - Short Papers (INTERMAG Short Papers) pp. 1 - 2
• Main Authors: Ross, Andrew, Leroux, Nathan, de Riz, Arnaud, Markovic, Danijela, Sanz-Hernandez, Dedalo, Trastoy, Juan, Bortolotti, Paolo, Querlioz, Damien, Martins, Leandro, Benetti, Luana, Claro, Marcel S., Anacleto, Pedro, Schulman, Alejandro, Taris, Thierry, Begueret, Jean-Baptiste, Saighi, Sylvain, Jenkins, Alex S., Ferreira, Ricardo, Vincent, Adrien F., Mizrahi, Alice, Grollier, Julie
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2023
• Subjects: Biological neural networks; Hardware; Junctions; magnetic tunnel junctions; Nanoscale devices; neural networks; neuromorphic computing; Neurons; Nonhomogeneous media; Radio frequency; radio-frequency multiplexing; Spintronics
• DOI: 10.1109/INTERMAGShortPapers58606.2023.10228503

Spintronic nano-synapses and nano-neurons are complex cognitive devices that have high accuracy due to their reproducible and controllable magnetization dynamics. They have the potential to revolutionize artificial intelligence hardware, but currently, there is no scalable way to connect them in multilayers. We show that magnetic tunnel junctions can be used to connect spintronics components in multilayer neural networks, allowing them to function as both synapses and neurons. We build a two-layer hardware spintronic neural network using nine magnetic tunnel junctions and found that it could classify nonlinearly-separable RF inputs with an accuracy of 97.7%. We also demonstrate that a larger network of these junctions could identify drones from their RF transmissions while consuming minimal power. This study lays the foundation for the development of deep, dynamical, spintronic neural networks [1].