Academic Journal
Method of construction of protected data transmission channels using modified neural network ; Метод побудови захищених каналів передачі даних з використанням модифікованої нейронної мережі
| العنوان: | Method of construction of protected data transmission channels using modified neural network ; Метод побудови захищених каналів передачі даних з використанням модифікованої нейронної мережі |
|---|---|
| المؤلفون: | Kal’chuk, Inna, Laptieva, Tetiana, Lukova-Chuiko, Nataliia, Kharkevych, Yurii |
| المصدر: | Collection "Information Technology and Security"; Vol. 9 No. 2 (2021); 232-243 ; Сборник "Information Technology and Security"; Том 9 № 2 (2021); 232-243 ; Information Technology and Security; Том 9 № 2 (2021); 232-243 ; 2518-1033 ; 2411-1031 |
| بيانات النشر: | ISCIP Igor Sikorsky Kyiv Polytechnic Institute |
| سنة النشر: | 2021 |
| مصطلحات موضوعية: | захищена мережа, передача даних, генетичний алгоритм, надійність, нейронна мережа, secure networks, data transmission, algorithm, reliability, neural network |
| الوصف: | Modern society is increasingly dependent on the quality of modern information and telecommunications services. An important indicator of the quality of such services is the security level of services provided. Therefore, the development of methods for constructing secure routes in information networks is an urgent scientific task. This article is devoted to solving this problem. The article considers the method of building secure routes in information networks. Traditional neural networks cannot provide modern capabilities for displaying secure networks, which are most important for data transmission analysis today. Therefore, Sigma-Pi-Sigma neural networks are a good tool for this operation due to their simple architecture. Applying an integrated approach to neural network learning that uses Sigma-Pi-Sigma neurons helps complete tasks in a short period. Neural networks need to find a solution to the problem. The Sigma-Pi-Sigma neural network model is used to accurately interpret the variable probe signal. The scientific novelty of the method is the construction of secure routes in information networks, it is a successful combination of the advantages of the radial basis and sigmoid activation functions. The gradient learning algorithm allows you to adjust the synaptic weights of the network in real-time with a given accuracy. The high learning speed and universal approximation properties of the proposed network are of practical importance; they will be especially useful when processing multidimensional vector argument functions. Future research will include the development of a Sigma-Pi-Sigma network without using the direct production procedure for hidden layer input vectors. To take advantage of existing methods, rectangular Fourier series matrix summation methods are used, which were not previously presented in similar methods. The efficiency of these methods for the study of secure data transmission is twice as high as triangular methods, which increases the probability of reliable data transmission by 15 ... |
| نوع الوثيقة: | article in journal/newspaper |
| وصف الملف: | application/pdf |
| اللغة: | Ukrainian |
| Relation: | http://its.iszzi.kpi.ua/article/view/250077/251244; http://its.iszzi.kpi.ua/article/view/250077 |
| DOI: | 10.20535/2411-1031.2021.9.2.250077 |
| الاتاحة: | http://its.iszzi.kpi.ua/article/view/250077 https://doi.org/10.20535/2411-1031.2021.9.2.250077 |
| Rights: | Copyright (c) 2021 Collection "Information Technology and Security" ; http://creativecommons.org/licenses/by/4.0 |
| رقم الانضمام: | edsbas.93440497 |
| قاعدة البيانات: | BASE |
| DOI: | 10.20535/2411-1031.2021.9.2.250077 |
|---|