Tunable superconducting neurons for networks based on radial basis functions

Andrey E Schegolev; Nikolay V Klenov; Sergey V Bakurskiy; Igor I Soloviev; Mikhail Yu Kupriyanov; Maxim V Tereshonok; Anatoli S Sidorenko

doi:10.3762/bjnano.13.37

Tunable superconducting neurons for networks based on radial basis functions

Beilstein J Nanotechnol. 2022 May 18:13:444-454. doi: 10.3762/bjnano.13.37. eCollection 2022.

Authors

Andrey E Schegolev^{1

2}, Nikolay V Klenov^{3

4}, Sergey V Bakurskiy^{1

5}, Igor I Soloviev¹, Mikhail Yu Kupriyanov¹, Maxim V Tereshonok², Anatoli S Sidorenko^{6

7}

Affiliations

¹ Skobeltsyn Institute of Nuclear Physics, Lomonosov Moscow State University, 119991 Moscow, Russia.
² Moscow Technical University of Communication and Informatics (MTUCI), 111024 Moscow, Russia.
³ Faculty of Physics, Lomonosov Moscow State University, 119991 Moscow, Russia.
⁴ Lobachevsky State University of Nizhni Novgorod Faculty of Physics, 603950 Nizhny Novgorod, Russia.
⁵ Dukhov All-Russia Research Institute of Automatics, 101000 Moscow, Russia.
⁶ Institute of Electronic Engineering and Nanotechnologies ASM, MD2028 Kishinev, Moldova.
⁷ Laboratory of Functional Nanostructures, Orel State University named after I.S. Turgenev, 302026, Orel, Russia.

Abstract

The hardware implementation of signal microprocessors based on superconducting technologies seems relevant for a number of niche tasks where performance and energy efficiency are critically important. In this paper, we consider the basic elements for superconducting neural networks on radial basis functions. We examine the static and dynamic activation functions of the proposed neuron. Special attention is paid to tuning the activation functions to a Gaussian form with relatively large amplitude. For the practical implementation of the required tunability, we proposed and investigated heterostructures designed for the implementation of adjustable inductors that consist of superconducting, ferromagnetic, and normal layers.

Keywords: Josephson circuits; networks on radial basis functions; radial basis functions (RBFs); spintronics; superconducting electronics; superconducting neural network.

Grants and funding

G-neuron and tunable inductance were developed with the support of the Russian Science Foundation (project no. 20-69-47013). The numerical simulations were supported within the framework of the strategic academic leadership program of UNN.