Magnetic Elements for Neuromorphic Computing

Abstract

Neuromorphic computing is assumed to be significantly more energy efficient than, and at the same time expected to outperform, conventional computers in several applications, such as data classification, since it overcomes the so-called von Neumann bottleneck. Artificial synapses and neurons can be implemented into conventional hardware using new software, but also be created by diverse spintronic devices and other elements to completely avoid the disadvantages of recent hardware architecture. Here, we report on diverse approaches to implement neuromorphic functionalities in novel hardware using magnetic elements, published during the last years. Magnetic elements play an important role in neuromorphic computing. While other approaches, such as optical and conductive elements, are also under investigation in many groups, magnetic nanostructures and generally magnetic materials offer large advantages, especially in terms of data storage, but they can also unambiguously be used for data transport, e.g., by propagation of skyrmions or domain walls. This review underlines the possible applications of magnetic materials and nanostructures in neuromorphic systems.

Neuro-Inspired Signal Processing in Ferromagnetic Nanofibers

“…This, on the other hand, is the basis for the common process of adding up signals. For this, it is necessary that not only "learning" is realized, but also "forgetting"; i.e., if a certain stimulus value (here named in this way to avoid confusion with the threshold values defined before) is not reached after a certain time, the sum of the signals is set back to its original value (here zero) and summing up starts again [48]. This process can be realized, e.g., by a mono-domain magnetic tunnel junction in which the input stimuli frequency must be high enough to allow for crossing the energy barrier that separates two stable states [49].…”

Stochasticvs.deterministic magnetic coding in designed cylindrical nanowires for 3D magnetic networks

“…following the idea of a memristor with the possibility of reversible changes. [23,24] For stochastic computing, [50] each nanowire can be set as a random sequence of bits. As is necessary for this application, we have a nonlinear system presenting weak correlations for switching.…”

“…In particular, multi-segmented nanowires [18,19,20] can provide active channels for domain wall pinning, spin wave manipulation, complex interactions in the collective magnetic reversal or thermal gradients accumulation. Other recent intriguing proposals are related to the use of 3D interconnecting arrays of magnetic nanowires for energy storage [21] and braininspired computing, [22,23,24] where 3D networks of magnetic NWs mimic the neurons and the constrictions between NWs the neural synapse. Finally, they can be also used for medical sensing and other healthcare applications.…”

Chemical and magnetic order in mass-selected large FeRh nanomagnets embedded in a carbon matrix

“…Exploring the magnetic properties of nanostructures continues to attract significant attention thanks in part to potential applications in spintronics and neuromorphic computing [1] . However, for single-domain nanoparticles, where the magnetic anisotropy energy (MAE) is in competition with the thermal activation energy, the superparamagnetic (SP) behaviour is not compatible with data storage at ambient temperature.…”