OSOleh Suprun

Oleh Suprun

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Oleh Suprun has a total of 30 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    ГІБРИДНІ АРХІТЕКТУРИ ШТУЧНОГО ІНТЕЛЕКТУ ДЛЯ ПРОГНОЗУВАННЯ ПОВЕДІНКОВИХ АНОМАЛІЙ У МЕРЕЖЕВОМУ ТРАФІКУ

    by О. М. Супрун, Я. Я. Кравчук, О. О. Супрун

    Актуальність дослідження зумовлена стрімким зростанням обсягів і різноманітності мережевого трафіку, ускладненням архітектур сучасних розподілених мереж та зростанням частоти складних кіберінцидентів, що проявляються у формі поведінкових аномалій. За таких умов традиційні сигнатурні та окремі модельні підходи аналізу трафіку не забезпечують достатньої точності, стійкості й завчасності реагування, що зумовлює потребу в застосуванні комплексних інтелектуальних підходів до прогнозування аномальної мережевої поведінки. Метою статті є обґрунтування підходів і підвищення ефективності прогнозування поведінкових аномалій у мережевому трафіку на основі використання гібридних архітектур штучного інтелекту (далі – ШІ), здатних забезпечити адаптивне, стійке та випереджальне виявлення аномальної мережевої поведінки в умовах динамічних і гетерогенних мережевих середовищ. Методи дослідження ґрунтуються на теоретичному аналізі сучасних наукових джерел у сфері ШІ та кібербезпеки, системному підході до дослідження мережевих процесів, логічному узагальненні результатів і порівняльному аналізі підходів до прогнозування аномалій у мережевому трафіку із застосуванням різних класів інтелектуальних моделей. Результати дослідження свідчать, що ізольоване застосування окремих моделей ШІ є обмежено ефективним у завданнях раннього прогнозування поведінкових аномалій, тоді як їхнє архітектурне поєднання дає змогу комплексно враховувати часові, структурні та контекстні характеристики трафіку. З’ясовано, що гібридні архітектури ШІ підвищують точність і стійкість прогнозів, зменшують чутливість моделей до дрейфу даних і створюють часову перевагу для превентивного реагування. Ідентифіковано основні науково-практичні проблеми впровадження таких рішень, пов’язані з обмеженою якістю навчальних даних, масштабованістю систем, інтерпретованістю результатів і ризиками цілеспрямованого спотворення моделей. Висновки. Гібридні архітектури ШІ доцільно розглядати як системну основу для інтелектуального аналізу мережевого трафіку, яка забезпечує перехід від реактивного виявлення інцидентів до проактивного, прогнозноорієнтованого управління кібербезпекою. Перспективи подальших досліджень пов’язані з поглибленням теоретичних засад побудови гібридних архітектур ШІ, розробленням узгоджених критеріїв оцінювання якості раннього прогнозування та дослідженням підходів до підвищення стійкості інтелектуальних систем аналізу трафіку в розподілених і критичних мережевих інфраструктурах.

  • FATTY ACID COMPOSITION OF GRAIN OF EINKORN AND ITS RELATIVES

    by Liana Relina, Oleh Suprun, Roman Bohuslavskyi, Liudmyla Vecherska, Olha Antsyferova, Nina Ilchenko, Valeriia Kolomatska, Liubov Kobyzeva

    Einkorn (Triticum monococcum L.) has a number of benefits attributed to the quality of its grain. Nevertheless, einkorn oil is little studied. Given a renewed interest in this crop, the purpose of the study was to investigate the lipid contents and fatty acid composition of oil from diploid wheat grain. Oil was extracted from dried whole wheat kernels by Soxhlet procedure. Fatty acid composition was determined by gas chromatography. The oil yield from the diploid wheat grain varied from 2.830.27% dry basis in Triticum monococcum var. sofianum Stranski to 4.460.49% in Triticum sinskayae A.Filat. et Kurk. Six major fatty acids were detected in all the wheat species under investigation. They are ranked in order of decreasing levels as follows: linoleic > oleic > palmitic > linolenic > stearic > palmitoleic. Ploidy doubling brought no significant alterations in fatty acid composition of T. monococcum grain. T. monococcum var. sofianum had the most beneficial unsaturated/saturated ratio (5.3) and the lowest the ratio of omega-6/omega-3 (9:1). Triticum boeoticum Boiss. (wild progenitor of einkorn) was inferior to domestic diploid wheat in terms of unsaturated fatty acid amounts, despite the very high total oil content (4.190.48%). As wheat oil is used in the cosmetics industry and given the surprisingly high oil yields from the diploid wheat grain, the collection accessions are worth considering through the lens of this trend in wheat products. However, none of the studied accessions can be recommended as breakthrough advantageous because of the ratios of omega-6/omega-3 of not lower than 9:1.


  • Information Transmission Protection Using Linguistic Steganography With Arithmetic Encoding And Decoding Approach

    by Maksym Ivasenko, Oleh Suprun

    Security of Internet of Things and Internet space in general is one of the main areas of research in the field of information technology today and steganography - one of the best methods of message encryption and transmission. Steganography algorithms are used to hide data in cover media, such as text, images, videos, etc., as cover media and change the structure and features of a text message so that it is not identified by human eyes and machines. However, the use of text media as accompanying text is relatively difficult compared to other methods of information coverage due to the lack of available redundant information in the text data. At the same time, text data is being transmitted the fastest and most frequently, so it is ideal for quickly transmitting short important messagesю In this research, an overview of existing approaches to textual steganography and our approaches to steganography methods are presented. New approaches to text steganography, which easily hide the text-coverage and also take less time to encrypt, are presented. A new method of steganography is shown and tested; a software implementation of this method is created. The advantages of using a new method of steganography are substantiated. It is confirmed that this algorithm is faster when working with large amounts of data, comparing to existing algorithms. At the same time, it allows to choose different parameters and settings, increasing its protective potential.


  • Forecasting and Analysis of Online Orders Rejections Using Data Mining Algorithms and Penalty Function

    by Oleh Suprun, Nina Klimenkova, Maksym Melnyk, Sonya Lavriy

    The direction of products and food delivery from online stores and similar transportation services have become especially relevant in recent years in connection with the rapid informatization of society, and especially because of quarantine restrictions due to the worldwide pandemic. However, along with increasing demand, online stores and individual couriers often incur significant financial losses due to customer rejection of orders, disruption in preparation, or other resignation issues that are independent of the company. Therefore, the article presents a way to partially reduce financial and time losses by analyzing preliminary order statistics, identifying potential dependencies and, consequently, predicting future customer rejection based on order details. Given the large number of statistics, it is advisable to use the tools and paradigms of data mining, two different approaches are considered and compared. At the same time there is a problem of false positive conclusion of the model, due to which the system ignores the profitable orders. To reduce such losses, the use of the method of penalty functions is proposed, which allows to assess the model risks and possible benefits. The described approaches were tested on a large real sample, and the corresponding results are presented, which demonstrate the effectiveness of the approach.


  • Methods of Designing Adaptive Systems of Multilevel Monitoring and Diagnosis for Recognition and Forecasting of Technological Condition of Complex Technical Objects

    by Nadiia Marchenko, Olena Nechyporuk, Oksana Martynova, Oleh Suprun, Maksym Melnyk

    The complex problem of building a multi-level system for monitoring and diagnosing of complex industrial facilities as integrated systems based on the principles of self-organization of complex systems is presented. This problem is especially relevant for large companies, industrial systems and strategically important objects, since even a minor mistake may lead to great financial loses or even technological catastrophe. Therefore, a reliable system for problems forecasting is required. The proposed two-level system of monitoring and diagnosing the state of complex industrial facilities is proposed, which differs from analogues by the ability to automatically select the optimal modes of operation of subsystems when changing the characteristics of the incoming flow of applications. This improvement allows not only to consider the state of the system, but to predict different future changes, according to expert’s predictions.


  • Open Access

    Devising a method of protection against zero-day attacks based on an analytical model of changing the state of the network sandbox

    by Serhii Buchyk, Oleksandr Yudin, Ruslana Ziubina, Ivan Bondarenko, Oleh Suprun

    This paper reports a method of protection against zero-day attacks using SandBox technology based on the developed analytical model with a probabilistic ranking of information system states. The model takes into consideration the conditions of a priori uncertainty regarding the parameters of the destructive flow on the system, accounting for the typical procedures of the network SandBox. The proposed model of information system states makes it possible to analyze and track all possible states, as well as assess the level of security in these states, and the probability of transitions into them. Thus, it is possible to identify the most dangerous ones and track the activities that caused the corresponding changes. The fundamental difference between this model and standard approaches is the weight coefficients that characterize not the intensity of random events but the intensity of transitions between states. Direct implementation and application of the proposed analytical model involved the technology of multilevel network "SandBoxes". The difference from other popular anti-virus tools is the use of a priori mathematical threat assessment, which makes it possible to detect influences that are not considered threats by classical systems until the moment of harm to the system. The combination with standard security tools makes it possible to separately analyze files that are too large in size, whether they enter the system not through a common gateway controlled by the network "SandBox" but from the external media of end-users. The implementation of the developed analytical model has made it possible to improve the level of protection of the corporate network by 15 %, based on the number of detected threats. This difference is explained by the inability of classical software to detect new threats if they are not already listed in the database of the program, and their activity is not trivial


  • Research of Features and Possibilities of Modern Real Time Video Services

    by Anton Sorokun, Oleh Suprun, Viktor Matviichuk, Serhii Voskoboinikov, Yurii Babenko

    Modern trends in telecommunications are aimed at the applying of science-intensive technologies to the various classes of video information services. Such services gained high popularity during last decade in different spheres of human life, like entertainment segment, and using video-conferences for work or study. At the same time, due to the technical characteristics and limitations of modern communication channels, the requirements needed to ensure high quality of video services that use modern information and communication systems can not be fully provided. Reducing time delays in the delivery of digital video presentations by services that use infocommunication technologies to improve the video quality is an important scientific and applied problam. The method for solving this problem is presented in the article. It is shown and proven that an imbalance between the technical capabilities of modern information and communication systems (ICS) like the data transfer speed, and the rapidly growing requirements for high-quality real time video services is very large, especially for mobile segment and wireless connections. An information model to represent the set of influential elements of video coherence domains in the form of structural-positional numbers with the presence of properties of global and local inequality of adjacent elements is created. A model for evaluating of the informativeness of video areas based on the identification and formation of the set of influential coherence areas is built.


  • Personal Documents Identification System Development Using Neural Network

    by Svitlana Popereshnyak, Oleh Suprun, Tadeusz Wieckowski

    The problem of analyzing and classification of digital images, that contain the documents, is very common for finance and economic spheres, for example, for bank and government agencies. Analyzing the documents differs from photos classification since the documents very often have low contrast and are similar to each other. For the personal documents classification, the structure of the convolutional neural network has been designed and developed. To achieve high accuracy of the recognition, that is very important, working with documents, the optimal parameters for the designed network were determined experimentally. The problems, that occur using such system, are mentioned. A method, used to avoid overfitting of the network, is described. The experimental network has been approved by practical experiments, their results are presented.


  • IoT application testing features based on the modelling network

    by Svitlana Popereshnyak, Oleh Suprun, Tadeusz Wieckowski

    In the article the features of testing devices and IoT applications are investigated. To make the problem clearer, the basic aspects of IoT applications architecture are considered. A set of requirements to the testing is presented, which must be considered when planning and organizing the testing process for IoT applications. The difficulties that arise during the testing phase of IoT applications and systems are described, same as mane differences from the classic application testing. As an approbation of the proposed test specifications, a full-scale experiment was conducted to test a fragment of the communication network in order to select an application protocol for a typical IoT, according to the results the conclusions are made.


  • Intrusion detection method based on the sensory traps system

    by Svitlana Popereshnyak, Oleh Suprun, Tadeusz Wieckowski

    In this paper, intrusion detection system based on touch traps is presented. Aim of the work - investigate and design a software product for monitoring intruders in various systems for further analysis and prevention of repeated attacks. The work is designed for intrusion detection system that includes procedures for analyzing network honeypots and procedure for systematization received data. The results may be used for: Improving intrusion detection system; Providing users of intrusion detection system tool for operational monitoring of network honeypots; Providing additional tools for systematization of input data. The efficiency of the above-mentioned procedures verified by computer experiment.