Implementation of new superconducting neural circuits using coupled SQUIDs

Mizugaki, Yoshinao; Nakajima, Kensuke; Sawada, Yasuji; Yamashita, Tsutomu

doi:10.1109/77.273058

Cited by 33 publications

(14 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further, introducing the sum Σi c = i c1 + i c2 and difference ∆i c = i c1 − i c2 of the critical currents, and taking (1) into account one can represent (7) in the following form:…”

Section: B Artificial Synapsementioning

confidence: 99%

Adiabatic superconducting artificial neural network: Basic cells

Soloviev

Schegolev

Klenov

et al. 2018

Journal of Applied Physics

View full text Add to dashboard Cite

We consider adiabatic superconducting cells operating as an artificial neuron and synapse of a multilayer perceptron (MLP). Their compact circuits contain just one and two Josephson junctions, respectively. While the signal is represented as magnetic flux, the proposed cells are inherently nonlinear and close-to-linear magnetic flux transformers. The neuron is capable of providing a one-shot calculation of sigmoid and hyperbolic tangent activation functions most commonly used in MLP. The synapse features by both positive and negative signal transfer coefficients in the range ∼ (−0.5, 0.5). We briefly discuss implementation issues and further steps toward multilayer adiabatic superconducting artificial neural network which promises to be a compact and the most energy-efficient implementation of MLP.

show abstract

“…Further, introducing the sum Σi c = i c1 + i c2 and difference ∆i c = i c1 − i c2 of the critical currents, and taking (1) into account one can represent (7) in the following form:…”

Section: B Artificial Synapsementioning

confidence: 99%

Adiabatic superconducting artificial neural network: Basic cells

Soloviev

Schegolev

Klenov

et al. 2018

Journal of Applied Physics

View full text Add to dashboard Cite

show abstract

“…In semiclassical regime, the transition rate by quantum tunneling 17 is roughly given by replacing k B T with ℏω in Eq. (8). In this way, the above generation scheme is made available for any system with stochastic transition processes.…”

Section: Physical Basis Of Artificial Neuronsmentioning

confidence: 99%

“…In fact, ANN using Josephson elements has been proposed 5,6 in the early 90s, there have been successful implementations using different ways. [7][8][9][10][11] In the previous studies, different types of activation functions in artificial neurons are employed to construct the superconducting artificial neural networks. For example, the two-stage coupled SQUID with a cascade connection produced a step-like function 9 and the rapid single flux quantum (RSFQ) comparator based on the statistical transition provided the error function similar to sigmoid function 10 .…”

Section: Introductionmentioning

confidence: 99%

Theoretical basis of SQUID-based artificial neurons

Katayama

Fujii

Hatakenaka

2018

Journal of Applied Physics

View full text Add to dashboard Cite

The physical basis of an artificial neuron is studied using a model that is based on the stochastic transition between two states in a double well potential. It is shown that the stochastic transition model generates an energy-defined sigmoid function acting as an activation (or transfer) function in neurons. The model is also applied to circuit neurons using superconducting quantum interference devices (SQUIDs) in artificial neural networks.

show abstract

“…Our previous neuron size was 130pm x 110pm [7]. We have redesign the layout, and the neuron size has been reduced to 120pm x 95pm.…”

Section: A Neuronmentioning

confidence: 99%

“…A 2-bit neural-based A/D converter, which contained two coupled-SQUID neurons and one synapse resistor, was fabricated and operated as a network solving an optimization problem. We have also proposed a synapse of variable current source type, which can improve the performance of superconducting neural networks [7].…”

Section: : Introductionmentioning

confidence: 99%

Superconducting neural circuits using SQUIDs

Mizugaki¹,

Nakajima²,

Sawada³

et al. 1995

IEEE Trans. Appl. Supercond.

Self Cite

View full text Add to dashboard Cite

Superconducting implementation of neural circuits is presented. A neuron element is composed of a coupled-SQUID, which is a combination of a single-junction SQUID and a double-junction SQUID. A synapse element is composed of multiple double-junction SQUIDs. To demonstrate the abilities of these circuits experimentally, a 3-bit neuralbased A/D converter, which contains three neurons and three synapses, has been fabricated using a niobium integration technology. This is the prst successful implementation of both neuron and active synapse elements into a superconducting neural network.

show abstract

Implementation of new superconducting neural circuits using coupled SQUIDs

Cited by 33 publications

References 19 publications

Adiabatic superconducting artificial neural network: Basic cells

Adiabatic superconducting artificial neural network: Basic cells

Theoretical basis of SQUID-based artificial neurons

Superconducting neural circuits using SQUIDs

Contact Info

Product

Resources

About