HYBRID NEURAL NETWORK MODEL FOR PROBABILITY DISTRIBUTION IDENTIFICATION IN THE ASSESSMENT OF THE STATE OF NATURAL AND TECHNICAL OBJECTS
DOI:
https://doi.org/10.34185/1562-9945-5-166-2026-04Keywords:
hybrid neural network, probability distribution identification, empirical distribution function, statistical features, deep learning, mathematical statisticsAbstract
The aim of this study is to develop a hybrid neural network model for the automatic identification of the probability distribution through the combined use of the empirical distribution function and statistical characteristics of the sample. The research methodology integrates approaches from probability theory, mathematical statistics, deep learning, and computer simulation. The proposed architecture combines a one–dimensional convolutional neural network (1D CNN) for analyzing the shape of the empirical distribution with a multilayer perceptron (MLP) for processing a vector of 12 robust statistical features. A dataset of 60,000 synthetic samples with sizes ranging from 20 to 1000 observations was generated for model training and testing. The experimental results demonstrated a classification accuracy of 97.8% on an independent test set containing 18,000 samples, outperforming standalone CNN and MLP models as well as conventional statistical goodness–of–fit tests. The obtained results confirm the effectiveness of the proposed hybrid neural network approach for the automatic identification the probability distributions under conditions of limited data availability.
References
1. Tryfonova, K. (2025). Detection of hidden data in images based on the Kolmogorov–Smirnov criterion. In Information Control Systems and Technologies (ICST–Odesa–2025) (pp. 57–59). Odesa, Ukraine. Available at: https://icst–conf.com/2025.pdf
2. Tulyzyk, O., & Rudnichenko, M. (2025). Analysis of scheduling methods using genetic algorithms, simulated annealing, swarm intelligence, and neural networks. In Information Control Systems and Technologies (ICST–Odesa–2025) (pp. 204–206). Odesa, Ukraine. Available at: https://icst–conf.com/2025.pdf
3. Shvedov, D., Rudnichenko, M., & Petrov, I. (2025). Concept of a hybrid deep learning model for financial risk analysis. In Information Control Systems and Technologies (ICST–Odesa–2025) (pp. 209–211). Odesa, Ukraine. Available at: https://icst–conf.com/2025.pdf
4. Sudyn, A., Stsislo, L., Perekrest, A., & Onyshchenko, Y. (2025). Hybrid neural network model for targeted attack risk assessment in SCADA/ICS systems. Information Technologies and Society, 4(19), 162–169. https://doi.org/10.32689/maup.it.2025.4.25
5. Ivaniuk, I., & Hnatiuk, D. (2025). Study of vulnerabilities and protection in IoT networks using artificial intelligence. In Proceedings of Chernivtsi National University (pp. 158–159). Available at: https://www.chnu.edu.ua/media/leeld04z/2025–nniftkn_compressed.pdf
6. Abbasimehr, H., Paki, R., & Bahrini, A. (2021). Improving the performance of deep learn-ing models using statistical features. Mathematical Methods in the Applied Sciences, 49(5), 3496–3510. https://doi.org/10.1002/mma.7500
7. Fawaz, H. I., Lucas, B., Forestier, G., Pelletier, C., Schmidt, D. F., Weber, J., Webb, G. I., Idoumghar, L., Muller, P.–A., & Petitjean, F. (2020). InceptionTime: Finding AlexNet for time series classification. Data Mining and Knowledge Discovery, 34, 1936–1962. https://doi.org/10.1007/s10618–020–00710–y
8. Vanotchak, O. (2026). Development of a hybrid CNN–LSTM model for drilling complica-tion prediction. Herald of Khmelnytskyi National University, 363(2), 125–134. https://doi.org/10.31891/2307–5732–2026–363–16
9. Supruni, O. M., Ohirko, O. I., & Krymska, A. O. (2025). Development of hybrid neural network models for forecasting. Bulletin of Kherson National Technical University, 1(92), Part 2, 222–230. https://doi.org/10.35546/kntu2078–4481.2025.1.2.30
10. Klymenko, O., & Vovk, S. (2025). Comparative analysis of the normality of statistical criteria for samples of contaminated data. Journal of Rocket–Space Technology, 34(4), 97–104. https://doi.org/10.34185/1562–9945–5–160–2025–03
11. Klymenko, O. (2025). Sensitivity analysis of statistical criteria to pseudorandom variable generation parameters. Journal of Rocket–Space Technology, 34(2), 122–130. https://doi.org/10.15421/452521
12. Klymenko, O., & Malaichuk, V. (2024). Methods of assessing the conformity of a pseu-dorandom sequence to a random distribution in the problems of automation, cybersecurity, computer–integrated technologies and non–destructive testing. In Proceedings of the Interna-tional Scientific and Technical Conference "Information Technologies in Metallurgy and Me-chanical Engineering" (pp. 327–334). https://doi.org/10.34185/1991–7848.itmm.2024.01.062
13. Klymenko, O., Astakhov, D., & Klymenko, S. (2021). Theoretical studies of measure-ments using probability distribution functions. Journal of Rocket–Space Technology, 28(4), 152–158. https://doi.org/10.15421/452020
14. Motsnyi, F. V. (2018). Chi–square, Student's t, and Fisher–Snedecor statistical distribu-tions and their applications. Statistics of Ukraine, 80(1), 16–23. https://doi.org/10.31767/su.1(80).2018.01.02
Downloads
Published
Issue
Section
License
Copyright (c) 2026 System technologies
This work is licensed under a Creative Commons Attribution 4.0 International License.
