Neural network model of countering network cyber attacks using expert knowledge


Bapiyev I. Kamalova G. Yermukhambetova F. Khairullina A. Kassymova A.
15 July 2021Little Lion Scientific

Journal of Theoretical and Applied Information Technology
2021#99Issue 133179 - 3190 pp.

Research in the field of countering cyberattacks on network resources of information systems has shown that most modern neural network models are focused on learning using statistical data. Such models are not sufficiently adapted to recognize new types of network cyberattacks. To eliminate this drawback, it was proposed to form a training sample using expert knowledge presented in the form of production rules. It was determined that among the classical types of neural network models, the most suitable for such training is a probabilistic neural network. On the basis of this network, an original neural network model was created, its structure and software were developed. The use of the developed model makes it possible to increase the recognition efficiency and expand many types of network attacks, the characteristics of which are not presented in statistical data. Another important advantage of the developed model is the information content of the output signal, which is sufficient for flexible setting of protective measures.

Expert Knowledge , Model , Neural Network Model , Probabilistic Neural Network , Production Rule , Recognition Of Cyber Attacks

Text of the article Перейти на текст статьи

West Kazakhstan Agrarian Technical University named after Zhangir Khan, West-Kazakhstan Region, Uralsk, Kazakhstan

West Kazakhstan Agrarian Technical University named after Zhangir Khan

10 лет помогаем публиковать статьи Международный издатель

Книга Публикация научной статьи Волощук 2026 Book Publication of a scientific article 2026