Adaptive Synaptogenesis Implemented on a Nanomagnetic Platform

The human brain functions very differently from artificial neural networks (ANN) and possesses unique features that are absent in ANN. An important one among them is "adaptive synaptogenesis" that modifies synaptic weights when needed to avoid catastrophic forgetting and promote lifelong learning. The key aspect of this algorithm is supervised Hebbian learning, where weight modifications in the neocortex driven by temporal coincidence are further accepted or vetoed by an added control mechanism from the hippocampus during the training cycle, to make distant synaptic connections highly sparse and strategic. In this work, we discuss various algorithmic aspects of adaptive synaptogenesis tailored to edge computing, demonstrate its function using simulations, and design nanomagnetic hardware accelerators for specific functions of synaptogenesis.

An Unconventional Ultra-Sub-Wavelength Receiving Nano-Antenna Activated by ac Spin Pumping and the ac Inverse Spin Hall Effect

We report an extreme sub-wavelength unconventional receiving antenna. It consists of an array of nanomagnets connected to heavy metal nanostrips. Incident electromagnetic (EM) radiation generates intrinsic and extrinsic spin waves in the nanomagnets, which pump spin into the heavy metal nanostrips at their own frequencies giving rise to a polychromatic alternating voltage across the latter owing to the ac inverse spin Hall effect. This implements a receiving nano-antenna. We demonstrate its operation at two different EM wave frequencies of 1.5 GHz and 2.4 GHz - the latter being the Bluetooth and Wi-Fi frequency. We measure the receiving gain at 2.4 GHz to be approximately -9 db. The free space radiated wavelength "lambda" at 2.4 GHz is 12.5 cm while the antenna area A is merely 160 micron^2, making the ratio A/lambda^2 = 0.97x10^-8. This antenna's receiving gain should be very poor because of the tiny size. Yet the measured gain is more than 4000 times larger than the theoretical limit for a conventional antenna of this size at this wavelength because of the unconventional operating principle.

Ternary Stochastic Neuron -- Implemented with a Single Strained Magnetostrictive Nanomagnet

Stochastic neurons are extremely efficient hardware for solving a large class of problems and usually come in two varieties -- "binary" where the neuronal statevaries randomly between two values of -1, +1 and "analog" where the neuronal state can randomly assume any value between -1 and +1. Both have their uses in neuromorphic computing and both can be implemented with low- or zero-energy-barrier nanomagnets whose random magnetization orientations in the presence of thermal noise encode the binary or analog state variables. In between these two classes is n-ary stochastic neurons, mainly ternary stochastic neurons (TSN) whose state randomly assumes one of three values (-1, 0, +1), which have proved to be efficient in pattern classification tasks such as recognizing handwritten digits from the MNIST data set or patterns from the CIFAR-10 data set. Here, we show how to implement a TSN with a zero-energy-barrier (shape isotropic) magnetostrictive nanomagnet subjected to uniaxial strain.

A Spintronic Nano-Antenna Activated by Spin Injection from a Three-Dimensional Topological Insulator

A charge current flowing through a three-dimensional topological insulator (3D-TI) can inject a spin current into a ferromagnet placed on the surface of the 3D-TI. Here, we report leveraging this mechanism to implement a nano-antenna that radiates an electromagnetic wave (1-10 GHz) into the surrounding medium efficiently despite being orders of magnitude smaller than the radiated free space wavelength. An alternating charge current of 1-10 GHz frequency is injected into a thin film of the 3D-TI Bi2Se3, resulting in the injection of an alternating spin current (of the same frequency) into a periodic array of cobalt nanomagnets deposited on the surface of the 3D-TI. This causes the magnetizations of the nanomagnets to oscillate in time and radiate electromagnetic waves in space, thereby implementing a nano-antenna. Because it is so much smaller than the free space wavelength, the nano-antenna is effectively a "point source" and yet it radiates anisotropically because of internal anisotropy. One can change the anisotropic radiation pattern by changing the direction of the injected alternating charge current, which implements beam steering. This would normally not have been possible in a conventional extreme sub-wavelength antenna.