S.V. Zimina – Ph.D. (Phys.-Math.), Associate Professor, Lobachevsky State University of Nizhny Novgorod

E-mail: zimina-sv@yandex.ru

Setting up artificial neural networks using iterative algorithms is accompanied by fluctuations in weight coefficients. These fluctuations occur due to the statistical relationship between the input signal vectors and the weight vectors of artificial neurons and lead to a deterioration in the quality of tuning of artificial neural networks.

The goal of the research is a statistical analysis of the operation of an artificial neural network that is configured using a fast recurrent algorithm with restrictions, taking into account fluctuations in the configurable weight coefficients.

Expressions of the correlation function and output power of an arbitrary neuron located in an arbitrary layer of an artificial neural network were calculated. It is shown that accounting for fluctuations leads to the appearance of additional terms in the statistical characteristics that distort the useful signal allocated by an artificial neural network.

Records of fluctuations of the weight vector when setting up artificial neural networks allows to optimize the performance of these networks in terms of "speed settings - tune", i.e. to choose at a given level of distortion fluctuations in the statistical characteristics of the allocated network of the signal the maximum speed setting of the network.

Зимина С.В. Характеристики искусственной нейронной сети с быстрым рекуррентным алгоритмом настройки с учетом флуктуаций весовых коэффициентов. Нейрокомпьютеры: разработка, применение. 2020. Т. 22. № 3. С. 36–44. DOI: 10.18127/j19998554-202003-04

1. Khaykin S. Neyronnyye seti: polnyy kurs. Izd. 2-e, ispr. Per. s angl. M.: OOO «I.D. Vi-liams». 2006. 1104s. (in Russian).
2. Galushkin A.I. Teoriya neyronnykh setey. Kn.1: Ucheb. posobiye dlya vuzov. Obshchaya red. A.I. Galushkina. M.: IPRZhR. 2000. 416 s. (in Russian).
3. Tatuzov A.L. Neyronnyye seti v zadachakh radiolokatsii. Kn.28. M.: Radiotekhnika. 2009. 432s. (in Russian).
4. Bershad N.J., Eweda E., Bermudez J.C.M. Stochastic analysis of an adaptive line enhancer/canceler with a cyclostationary input. IEEE Transactions on Signal Processing. 2016. V. 64. № 1. P. 104-119. DOI: 10.1109/TSP.2015.2486745
5. Jeong J.J., Kim S.H., Koo G., Kim S.W. Mean – square deviation analysis of multiband – struc-tured subband adaptive filter algorithm. IEEE Transactions on Signal Processing. 2016. V. 64. № 4. P. 985-994. DOI: 10.1109/TSP.2015.2498136
6. Eweda E., Bershad N.J. Stochastic analysis of the signed LMS algorithms for cyclostationary white gaussian inputs. IEEE Transactions on Signal Processing. 2017. V. 65. № 7. P. 1673-1684. DOI: 10.1109/TSP.2016.2646666
7. Li Z., Xia Y., Pei W., Wang K., Mandic D.P. An augmented nonlinear LMS for digital self – in-terference cancellation in full – duplex direct – conversion transceivers. IEEE Transactions on Signal Processing. 2018. V. 66. № 15. P. 4065-4078. DOI: 10.1109/TSP.2018.2846250
8. Zeng X., Yeung D.S. Sensitivity analysis of multilayer perceptron to input and weight perturba-tions. IEEE Transactions on Neural Networks. 2001. V. 12. № 6. P. 1358 - 1366. DOI: 10.1109/72.963772
9. Zimina S.V. Statisticheskiye kharakteristiki iskusstvennoy neyronnoy seti s diskret-nym gradiyentnym algoritmom nastroyki s uchetom fluktuatsiy vesovykh koeffitsiyentov. Neyrokompyutery: razrabotka i primeneniye. 2006. № 10. S. 9–15. (in Russian).
10. Zimina S.V. Vliyaniye fluktuatsiy vesovykh koeffitsiyentov na statisticheskiye kharakteristiki iskusstvennoy neyronnoy seti s algoritmom rekurrentnogo obrashcheniya vyboroch-noy otsenki korrelyatsionnoy matritsy vkhodnykh signalov. Neyrokompyutery: razrabotka, primeneniye. 2007. № 5. S. 3–7. (in Russian).
11. Litvinov O.S., Zimina S.V. Statisticheskiy analiz fluktuatsiy vesovykh koeffitsiyen-tov iskusstvennoy neyronnoy seti. nastraivayushcheysya po algoritmu Khebba. Neyrokompyutery: razrabotka, primeneniye. 2009. № 3. S. 33–43. (in Russian).
12. Zimina S.V. Fluktuatsii vesovykh koeffitsiyentov v iskusstvennoy neyronnoy seti s al-goritmom Khebba. Neyrokompyutery: razrabotka, primeneniye. 2013. № 4. S. 3 – 8. (in Russian).
13. Pupkov K.A., Kapalin V.I., Yushchenko A.S. Funktsionalnyye ryady v teorii nelineynykh sistem. M.: Nauka. 1976. (in Russian).
14. Zimina S.V. Fluktuatsii vesovogo vektora v adaptivnykh antennykh reshetkakh s neliney-noy funktsiyey v tsepi obratnoy svyazi. nastraivayushchikhsya po algoritmu rekurrentnogo obrashcheniya vyborochnoy otsenki korrelyatsionnoy matritsy vkhodnykh signalov. Izv. VU-Zov. Radiofizika. 2006. T. 49. № 2. S. 164-173. (in Russian).
15. Kruglov V.V., Borisov V.V. Iskusstvennyye neyronnyye seti. Teoriya i praktika. Izd. 2-e. M.: Goryachaya liniya – Telekom. 2002. 382 s. (in Russian).