Adiabatic superconducting cells for ultra-low-power artificial neural networks

Andrey E Schegolev; Nikolay V Klenov; Igor I Soloviev; Maxim V Tereshonok

doi:10.3762/bjnano.7.130

Adiabatic superconducting cells for ultra-low-power artificial neural networks

Beilstein J Nanotechnol. 2016 Oct 5:7:1397-1403. doi: 10.3762/bjnano.7.130. eCollection 2016.

Authors

Andrey E Schegolev¹, Nikolay V Klenov², Igor I Soloviev³, Maxim V Tereshonok⁴

Affiliations

¹ Lomonosov Moscow State University, Physics Faculty, Moscow, 119991, Russia; Moscow Technical University of Communications and Informatics (MTUCI), Moscow, 111024, Russia.
² Lomonosov Moscow State University, Physics Faculty, Moscow, 119991, Russia; Moscow Technical University of Communications and Informatics (MTUCI), Moscow, 111024, Russia; Lomonosov Moscow State University Skobeltsyn Institute of Nuclear Physics, Moscow, 119991, Russia; N.L. Dukhov All-Russia Research Institute of Automatics, 127055, Moscow, Russia; Moscow Institute of Physics and Technology, State University, Dolgoprudniy, Moscow region, Russia.
³ Lomonosov Moscow State University Skobeltsyn Institute of Nuclear Physics, Moscow, 119991, Russia; Moscow Institute of Physics and Technology, State University, Dolgoprudniy, Moscow region, Russia; Lukin Scientific Research Institute of Physical Problems, Zelenograd, Moscow, 124460, Russia.
⁴ Moscow Technical University of Communications and Informatics (MTUCI), Moscow, 111024, Russia.

Abstract

We propose the concept of using superconducting quantum interferometers for the implementation of neural network algorithms with extremely low power dissipation. These adiabatic elements are Josephson cells with sigmoid- and Gaussian-like activation functions. We optimize their parameters for application in three-layer perceptron and radial basis function networks.

Keywords: Josephson effect; adiabatic superconductor cells; artificial neural networks; energy efficiency; superconductivity.