Radio-Frequency Multiply-and-Accumulate Operations with Spintronic Synapses

Abstract

Exploiting the physics of nanoelectronic devices is a major lead for implementing compact, fast, and energy efficient artificial intelligence. In this work, we propose an original road in this direction, where assemblies of spintronic resonators used as artificial synapses can classify analogue radio-frequency signals directly without digitalization. The resonators convert the radio-frequency input signals into direct voltages through the spin-diode effect. In the process, they multiply the input signals by a synaptic weight, which depends on their resonance frequency. We demonstrate through physical simulations with parameters extracted from experimental devices that frequency-multiplexed assemblies of resonators implement the cornerstone operation of artificial neural networks, the Multiply-And-Accumulate (MAC), directly on microwave inputs. The results show that even with a non-ideal realistic model, the outputs obtained with our architecture remain comparable to that of a traditional MAC operation. Using a conventional machine learning framework augmented with equations describing the physics of spintronic resonators, we train a single layer neural network to classify radio-frequency signals encoding 8x8 pixel handwritten digits pictures. The spintronic neural network recognizes the digits with an accuracy of 99.96 %, equivalent to purely software neural networks. This MAC implementation offers a promising solution for fast, low-power radio-frequency classification applications, and a new building block for spintronic deep neural networks.Presently, the most promising Artificial Intelligence algorithms are based on deep neural networks [10], which contain several layers of artificial neurons, each of them linked by synaptic connections: in each layer of an artificial neural network, the neuron signals are multiplied by synaptic weights, summed and injected into a neuron of the following layer (see Fig. 1a). This elementary operation is called Multiply-And-Accumulate (MAC). In a computer using the von Neumann architecture, weight multiplications and sums are performed by processing units, whereas synaptic weight values are stored in spatially separated memory units. In such architecture, the data flow between the processing and memory units induces a slowdown and excess energy consumption [11] that can be avoided by implementing the MAC operation in hardware, using in situ memory devices emulating neurons and synapses [12][13][14][15][16][17]. Neurons that take dc inputs and convert them to microwave signals have been demonstrated using spintronic nano-oscillators [18][19][20][21][22][23] and CMOS ring oscillators [24,25]. However, to this day there is no demonstration of tunable artificial synapses that directly perform MAC operations on microwave signals.

A quantum material spintronic resonator

“…Further, dc current can be used to modify the nanoconstriction's response, "in-situ" by local heating. These characteristics can be useful as spintronic resonators synapses, as discussed in 16,17 . In QM-SRs a dc current can adjust a synaptic "weight. "…”

“…Further, dc current can be used to modify the nanoconstriction’s response, “in-situ” by local heating. These characteristics can be useful as spintronic resonators synapses, as discussed in 16 , 17 . In QM-SRs a dc current can adjust a synaptic “weight.” Finally, the effects could be enhanced by using ferromagnetic materials with larger spin-phonon coupling and magnetoelastic effects, e.g.…”

“…Such spintronic resonators are nonlinear, and tunable in phase, amplitude and frequency 12,13 . This makes them of great interest in neuromorphic computing where their response to external perturbations, phase locking and mutual synchronization can be exploited [14][15][16][17] .…”

Non-volatile artificial synapse based on a vortex nano-oscillator

“…In recent years, spintronic devices have also been proposed for the hardware implementation of neuromorphic computing systems 11 – 13 . Spin torque nano-oscillators (STNOs) are among the most studied spintronic devices for use in ANNs 14 – 21 . An STNO consists of a spin valve 22 , 23 or, more often, an MgO-based magnetic tunnel junction (MTJ) 24 , 25 , combining a resistive output which depends on the magnetization state of a ferromagnetic layer with rich magnetization dynamics that can be manipulated by spin-polarized currents 26 , 27 .…”

“…Non-trivial tasks such as vowel recognition were achieved in a synchronization-based computing system 19 , where the recognition is performed based on the synchronization state between the input signal and a chain of four STNOs representing artificial neurons. More recently, STNOs have also been proposed to work as synapses in ANNs where, based on the spin diode effect, the synaptic weight depends on the frequency difference between the input RF signal and the resonator 20 , 21 . In this work, we propose the possibility of adding non-volatility to an STNO working as an artificial synapse, which in turn enables learning in an ANN using such devices.…”