Ross Andrew, Leroux Nathan, De Riz Arnaud, Marković Danijela, Sanz-Hernández Dédalo, Trastoy Juan, Bortolotti Paolo, Querlioz Damien, Martins Leandro, Benetti Luana, Claro Marcel S, Anacleto Pedro, Schulman Alejandro, Taris Thierry, Begueret Jean-Baptiste, Saïghi Sylvain, Jenkins Alex S, Ferreira Ricardo, Vincent Adrien F, Mizrahi Frank Alice, Grollier Julie
Unité Mixte de Physique CNRS/Thales, CNRS, Thales, Université Paris-Saclay, Palaiseau, France.
Centre de Nanosciences et de Nanotechnologies, Université Paris-Saclay, CNRS, Palaiseau, France.
Nat Nanotechnol. 2023 Nov;18(11):1273-1280. doi: 10.1038/s41565-023-01452-w. Epub 2023 Jul 27.
Spintronic nano-synapses and nano-neurons perform neural network operations with high accuracy thanks to their rich, reproducible and controllable magnetization dynamics. These dynamical nanodevices could transform artificial intelligence hardware, provided they implement state-of-the-art deep neural networks. However, there is today no scalable way to connect them in multilayers. Here we show that the flagship nano-components of spintronics, magnetic tunnel junctions, can be connected into multilayer neural networks where they implement both synapses and neurons thanks to their magnetization dynamics, and communicate by processing, transmitting and receiving radiofrequency signals. We build a hardware spintronic neural network composed of nine magnetic tunnel junctions connected in two layers, and show that it natively classifies nonlinearly separable radiofrequency inputs with an accuracy of 97.7%. Using physical simulations, we demonstrate that a large network of nanoscale junctions can achieve state-of-the-art identification of drones from their radiofrequency transmissions, without digitization and consuming only a few milliwatts, which constitutes a gain of several orders of magnitude in power consumption compared to currently used techniques. This study lays the foundation for deep, dynamical, spintronic neural networks.
自旋电子纳米突触和纳米神经元凭借其丰富、可重现且可控的磁化动力学，能够高精度地执行神经网络操作。这些动态纳米器件若能实现最先进的深度神经网络，便有望变革人工智能硬件。然而，目前尚无将它们连接成多层结构的可扩展方法。在此，我们展示了自旋电子学的旗舰纳米组件——磁性隧道结，可连接成多层神经网络，借助其磁化动力学实现突触和神经元功能，并通过处理、传输和接收射频信号进行通信。我们构建了一个由两层九个磁性隧道结连接而成的硬件自旋电子神经网络，并表明它能以97.7%的准确率对非线性可分射频输入进行原生分类。通过物理模拟，我们证明了一个由纳米级结组成的大型网络能够在不进行数字化且仅消耗几毫瓦功率的情况下，实现对无人机射频传输的最先进识别，与当前使用的技术相比，功耗降低了几个数量级。这项研究为深度、动态的自旋电子神经网络奠定了基础。
