Abstract. Superconducting digital systems are considered recently as
one of the most promising options for the
physical implementation of fast and energy-efficient
artificial neural networks. Possibility to combine advantages of Josephson active antenna structures, amplifiers
and analog-to-digital converters together
with neural network circuits in complexes for
cognitive signal processing is of particular interest.
We propose physical foundations for a new energy efficient implementation of artificial neural network algorithms. We
describe two neuron cells: Sigma-cell and
Gauss-cell with sigmoid- and Gaussian-like
activation functions respectively. We developed simple theory to optimize their transfer (flux-to-current) functions
for
application in three-layer perceptron and radial basis functions networks. Design of these cells is inspired by adiabatic
quantum flux parametron; the both have simple
topology and low energy consumption, working
in superconducting regime. Maintained similarity of designs allows to use well-developed adiabatic superconductor logic cells in interface circuits.
Key words: Josephson effect,
superconductivity, artificial neural network, neuron, activation function, RBF network,
Sigma-cell, S-cell, Gauss-cell, G-cell.
References
1.
Rosenblatt
F.X. Principles of Neurodynamics: Perceptrons and the Theory of Brain
Mechanisms. Spartan Books, Washington DC, 1961.
2.
Soloviev
I.I., Kornev
V.K., Sharafiev A.V., Klenov N. V., Mukhanov O. A. // IEEE Trans.
Appl. Supercond. 2013. V. 23.
N. 3. P. 1800405.
3.
Spietz
L., Irwin K., Aumentado J. // Appl.Phys.Lett. 2009. V. 95. P. 092505.
4.
Mukhanov
O. A. History of Superconductor Analog-to-Digital Converters In 100
Years of Superconductivity. Rogalla H., Kes P., Eds; London, 2011.
5.
Yan
Q., Li M., Chen F., Jiang T., Lou W., Hou T.Y. Lu C.-T. // IEEE Trans.
Wireless Commun. 2014. V. 13. P. 5893.
6.
Munjuluri
S., Garimella R.M. // Procedia Comput. Sci. 2015. V. 46. P. 1156.
7.
Farooqi
M., Tabassum S., Rehmani M., Saleem Y. J. // Network Comput. Appl. 2014.
V. 46. P. 1–16.
8.
Crotty
P., Schult D., Segall K. // Phys. Rev. E, 2010. V. 82. P. 011914.
9.
Harada
Y., Goto E. // IEEE Trans. on Magn. 1991. V. 27. P. 2863–2866.
10.
Rippert
E.D., Lomatch S. // IEEE Trans. Appl. Supercond. 1997. V. 7. P. 3442–3445.
11.
Chiarello
F., Carelli P., Castellano M.G., Torrioli G. // Supercond. Sci. Technol. 2013.
V. 26. P. 125009.
12.
Yamanashi
Y., Umeda K., Yoshikawa N. // IEEE Trans. Appl. Supercond. 2013. V. 23. P.
1701004.
13.
Onomi
T., Nakajima K. // J. Phys.: Conf. Series. 2014. V. 507. P. 042029.
14.
Likharev
K. // IEEE Trans. Magn. 1977. V. 13, 1. P. 242–244.
15.
Hosoya
M., Hioe W., Casas J., Kamikawai R., Harada Y., Wada Y., Nakane H., Suda
R., Goto E. // IEEE Trans. Appl. Supercond. 1991. V. 1. P. 77–89.
16.
Semenov
V.K., Danilov G.V., Averin, D.V. // IEEE Trans. Appl. Supercond. 2007.
V. 17. P. 455 - 461.
17.
Takeuchi
N., Yamanashi Y., Yoshikawa N. // Appl. Phys. Lett. 2013. V. 102. P. 052602.
18.
Nagasawa
S., Hashimoto Y., Numata H., Tahara S. // IEEE Trans. Appl. Supercond. 1995.
V. 5. P. 2447-2452.
19.
Nagasawa
S., Hinode K., Satoh T., Akaike H., Kitagawa Y., Hidaka M. // Physica C.
2004. V. 412-414. 2. P.
1429-1436.
20.
Takeuchi
N., Yamanashi Y., Yoshikawa N. // Supercond. Sci. Technol. 2015.
V. 28. P. 015003.
21.
Takeuchi
N., Yamanashi Y., Yoshikawa N. // J. Appl. Phys. 2015. V. 117. P. 173912.
22.
Kornev
V.K., Soloviev I.I., Klenov N.V., Mukhanov O.A. // Supercond. Sci. Tech. 2009. V.
22. P. 114011.
23.
Kornev
V.K., Soloviev I.I., Klenov N.V., Mukhanov O.A. // IEEE Trans. Appl. Supercond. 2007. V. 17, 2. P.
569-572.
24.
Adjemov
S.S., Klenov N.V., Tereshonok M.V., Chirov D.S. // Moscow University Phys.
Bull. 2015.
V. 70, 6. P. 448–456.
25.
Adjemov
S.S., Klenov N.V., Tereshonok M.V., Chirov D.S. // Programming and Computer
Soft. 2016. V. 42, 3. P. 121–128.
26.
S. S. Adjemov,
N. V. Klenov, M. V. Tereshonok,
D. S. Chirov // Moscow
University Phys. Bull.
2016. V.
71, 2. P. 174–179.