Reconfigurable classifier based on spin-torque-driven magnetization switching in electrically connected magnetic tunnel junctions

A promising branch of neuromorphic computing aims to perform cognitive operations in hardware, leveraging the physics of efficient and well-established nanodevices. This work presents a reconfigurable classifier, based on a network of magnetic tunnel junctions, that can learn to classify spoken vowels. In this task the hardware network surpasses multilayered software neural networks with the same number of trained parameters. These results, obtained using the same devices and working principle employed in industrial spin-transfer-torque magnetic random-access memory, constitute an important step toward the development of large-scale neuromorphic networks based on established technology.