Cybernetics And Systems Analysis logo
Editorial Board Announcements Abstracts Authors Archive
KIBERNETYKA TA SYSTEMNYI ANALIZ
International Theoretical Science Journal
-->

DOI 10.34229/KCA2522-9664.24.3.2
UDC 519.816(045)
А. Litvinenko1


1 National Aviation University, Kyiv, Ukraine

litvinen@nau.edu.ua

AN ALGEBRAIC METHOD FOR SYNTHESIZING ERROR-FREE
BINARY NEURAL NETWORK

Abstract. A mathematical model of the problem of calculating the weighting coefficients of a binary neural network is given. It is proved that in the case of step functions of neuron activation, this model is a system of linear inequalities, which is incompatible for most practical problems. A method of analyzing the system of inequalities is proposed, which allows calculating the values of the weighting coefficients and synthesizing the structure of the neural network, which ensures the absolute accuracy of the output signals. The algorithm and an implementation example are given.

Keywords: neural network, mathematical model, analysis, synthesis, error.


full text

REFERENCES

  1. Kazantsev A.V. Visual data processing and action control using binary neural network. Eight International Workshop on Image Analysis for Multimedia Interactive Services (WIAMIS’07) (Santorini, Greece, 6–8 June 2007). IEEE, 2007. Р. 23. https://doi.org//10.1109/WIAMIS.2007.90.

  2. Liang S., Sun R., Li Y., Srikant R. Understanding the loss surface of neural networks for binary classification. Proc. of the 35th International Conference on Machine Learning. 2018. P. 2835–2843. URL: https://proceedings.mlr.press/v80/liang18a/liang18a.pdf .

  3. Krinitskiy M., Verezemskaya P., Grashchenkov K., Tilinina N., Gulev S., Lazzara M. Deep convolutional neural networks capabilities for binary classification of polar mesocyclones in satellite mosaics. Atmosphere. 2018. Vol. 9, N 11. Article number 426. https://doi.org/10.3390/atmos9110426.

  4. Dunnmon J.A., Yi D., Langlotz C.P., Re C., Rubin D.L., Lungren M.P. Assessment of convolutional neural networks for automated classification of chest radiographs. Radiology. 2019. Vol. 290, N 2. Р. 537–544.

  5. Korolev S., Safiullin A., Belyaev M., Dodonova Y. Residual and plain convolutional neural networks for 3D brain mri classification. arXiv:1701.06643v1 [cs.CV] 23 Jan 2017. URL: https://arxiv.org/pdf/1701.06643.pdf .

  6. Menon A.K., Williamson R.C. The cost of fairness in binary classification. Proc. of the 1st Conference on Fairness, Accountability and Transparency. PMLR. 2018. Vol. 81. P. 107–118.

  7. Ferreyra-Ramirez A., Rodriguez-Martinez E., Aviles-Cruz C., Lopez-Saca F. Image retrieval system based on a binary auto-encoder and a convolutional neural network. IEEE Latin America Transactions. 2020. Vol. 18, Iss. 11. P. 1925–1932. https://doi.org//10.1109/TLA.2020.9398634.

  8. Lytvynenko O.E., Kucherov D.P., Hlybovets M.M. Decomposition method of calculating weight coefficients of a binary neural network. Kibernetyka ta systemnyi analiz. 2022. Vol. 58, N 6. P. 45–53.




© 2024 Kibernetika.org. All rights reserved.