Computer systems and information technologies

ANALYSIS OF DIFFUSION MODELS AND BIOMEDICAL IMAGE GENERATION TOOLS

Sergii KUZMIN, Oleh BEREZSKY — Thu, 26 Jun 2025 00:00:00 +0300

This study investigates the effective generation of realistic histopathological medical images through the fine-tuning of generative diffusion models, addressing critical needs in medical diagnostics, education, and research. High-quality synthetic histopathology images are essential for training medical professionals, augmenting limited datasets, and potentially enhancing diagnostic accuracy through machine learning applications. However, general-purpose image synthesis methods and limited annotated medical datasets pose significant challenges.

Four prominent fine-tuning methods—LoRA, DreamBooth, Textual Inversion, and HyperNetwork - were systematically evaluated using the Stable Diffusion 1.5 generative model. These methods were rigorously assessed using the balanced dataset with 664 images per each distinct tissue class: normal, serrated, adenocarcinoma, and adenoma tissues.

Quantitative evaluations employing Fréchet Inception Distance (FID), Precision, and Recall metrics revealed significant performance differences among the methods. HyperNetwork and DreamBooth consistently yielded superior image fidelity and diversity. Specifically, HyperNetwork achieved notably low FID scores (e.g., 77.27 for adenocarcinoma) accompanied by robust Precision and Recall results, demonstrating enhanced realism and variability. DreamBooth similarly exhibited strong performance, validating its practical utility. In contrast, Textual Inversion consistently produced the weakest outcomes, characterized by significantly higher FID scores (exceeding 158) and notably low Recall values, underscoring its inherent limitations for complex medical imaging applications.

Although these quantitative insights are valuable, traditional metrics alone may not comprehensively capture clinical applicability. Therefore, qualitative evaluation by medical professionals remains essential. Additionally, there is an urgent need for developing domain-specific evaluation metrics and fine-tuning techniques explicitly tailored for histopathology imaging. Such advancements hold the potential to significantly enhance synthetic image quality and expand their clinical and educational adoption.

SEMANTIC MODELS FOR WEB APPLICATION DESIGN

Iryna PIKH, Yulian MERENYCH — Thu, 26 Jun 2025 00:00:00 +0300

The development of web applications is often carried out within tight deadlines and under limited resources, which complicates adherence to the principles of high-quality design. The design of modern web applications requires consideration of a wide range of criteria that affect usability, accessibility, and the overall effectiveness of user interaction. Given the increasing complexity of digital interfaces, there is a growing need for the development of a theoretically grounded approach to evaluating design quality.

The aim of this study is to develop semantic models for formalizing the factors influencing the quality of web application design. The application of formalized models makes it possible to systematize expert knowledge, enhance the objectivity of decision-making, and ensure the reproducibility of results regardless of subjective factors.

This article identifies a set of factors that influence the quality of web application design. To achieve this, methods of expert evaluation and semantic modeling are employed. For the quantitative analysis of criterion significance, the rating scale method is applied, and indicators of variation and consistency of expert judgments are calculated. Three main categories of criteria are defined: ergonomics and cognitive principles of interaction; accessibility and inclusivity; and information architecture and visual design. Semantic models representing the relationships between factors within each category are constructed.

The proposed approach enables the formalization of expert knowledge and provides a foundation for the further automated evaluation of web application design quality using fuzzy logic methods. The developed semantic models can also be integrated into decision support systems for the design of digital products. Furthermore, they may serve as a basis for the development of educational materials and tools for auditing web application design in accordance with contemporary standards.

MONITORING SYSTEM FOR CRITICAL INFRASTRUCTURE OBJECTS BASED ON DIGITAL TWINS

Dmytro ANDRIEIEV, Oleksii LYHUN, Andriy DROZD — Thu, 26 Jun 2025 00:00:00 +0300

Critical infrastructures are fundamental to the seamless operation of modern societies, encompassing sectors such as energy, healthcare, transportation, and communications. Ensuring their reliability, performance, continuous operation, safety, maintenance, and protection is a national priority for countries worldwide.

The digital twins play a crucial role in critical infrastructure, as they enhance security, resilience, reliability, maintenance, continuity, and operational efficiency across all sectors. Among the benefits offered by digital twins are intelligent and autonomous decision-making, process optimization, improved traceability, interactive visualization, and real-time monitoring, analysis, and prediction. Furthermore, the study revealed that digital twins have the capability to bridge the gap between physical and virtual environments, can be used in combination with other technologies, and can be integrated into various contexts and industries.

The use of digital twins was explored as the foundation for developing a modern monitoring system for critical infrastructure facilities enables multi-level assessment of asset conditions in real time, ensuring precise threat detection, anomaly identification, and timely decision-making. Integration with artificial intelligence and big data technologies allows not only the collection and analysis of large volumes of information but also the creation of adaptive behavioral models for systems in emergency situations.

Special attention was given to the method of optimizing critical IT infrastructure using digital twins, which combines virtual modeling, predictive algorithms, and automated management. The proposed approach enhances the reliability of digital systems, minimizes downtime, optimizes maintenance costs, and strengthens cybersecurity. This system is especially relevant in the context of growing risks and increasing demands for the stability of strategically important infrastructure assets.

The application of digital twins for monitoring and optimizing critical infrastructure demonstrates considerable potential for improving its resilience, safety, and operational efficiency. The approaches discussed in the study confirm the relevance of implementing digital models as tools for timely risk identification, failure prediction, and informed decision-making. By integrating such technologies, organizations can reduce operational costs, minimize downtime, and improve the overall stability of infrastructure operations. Therefore, digital twins represent a vital step toward the digital transformation and modernization of mission-critical systems across various sectors.

INFORMATION TECHNOLOGY FOR ELECTROCARDIOGRAPHIC SIGNAL ANALYSIS BASED ON MATHEMATICAL MODELS OF TEMPORAL AND AMPLITUDE VARIABILITY

Lyubomyr MOSIY, Andriy SVERSTIUK — Thu, 26 Jun 2025 00:00:00 +0300

This paper presents an information technology for electrocardiographic signal analysis based on discrete mathematical models with temporal rhythm functions and amplitude variability of characteristic waves P, Q, R, S, T. A discrete mathematical model of the temporal rhythm function considering extreme amplitude values of ECS characteristic waves and an amplitude variability model have been developed for comprehensive analysis of morphological and rhythmic diagnostic features of cardiac signals. Experimental validation was conducted on ECS signals from patients with diagnoses: conditional norm and extrasystole. For patients with conditional norm, high stability of temporal intervals between ECS characteristic waves is observed with a mathematical expectation of 0.776 s for all wave types and minimal amplitude variability (mathematical expectation 0.00003-0.00064 mV, variance 0.00010-0.00022 mV). In patients with extrasystole, significant cardiac rhythm irregularity was detected with a decrease in mathematical expectation to 0.503-0.504 s (by 35%) and a three-order magnitude increase in variance (to 0.011-0.012 s) for temporal rhythm functions. The amplitude variability function demonstrated exponential growth of all statistical parameters: mathematical expectation increased to 0.070-0.452 mV (from 233 to 15067 times), variance reached extreme values of 78.44-719.20 mV (5-6 order magnitude increase), range varied within 46.2-122.9 mV (960 to 1500 times increase). The proposed discrete mathematical models successfully combine temporal rhythm functions considering extreme amplitude values of ECS characteristic waves with amplitude variability functions, enabling comprehensive assessment of both rhythmic and morphological ECS features. The models demonstrate high sensitivity to pathological changes in the cardiovascular system and expand the methodological foundation for developing information technology for expert analysis of morphological and rhythmic features of cardiac signals through integration with machine learning and artificial intelligence methods.

EDGE-NATIVE CABLE ACCESS NETWORK WITH UDP TERMINATION

Ivan IVANETS, Volodymyr OVSYAK, Oleksandr OVSYAK — Thu, 26 Jun 2025 00:00:00 +0300

This paper introduces a software-defined architectural model for modernizing upstream transport in hybrid fiber-coaxial networks through user-space UDP termination using the VPP HostStack. The approach aims to overcome performance limitations inherent in DOCSIS-based systems by relocating transport-layer processing from centralized cores to multi-access edge computing nodes co-located with Remote PHY Devices. Unlike centralized vCMTS deployments or full-fiber upgrades, this model enables enhanced throughput and session-level concurrency without modifying coaxial infrastructure, DOCSIS PHY signaling, or customer premises equipment. The system is built on FD.io’s Vector Packet Processing framework and validated through CI-integrated benchmarking using the CSIT testbed and iperf3 traffic generator. Test executions were performed on a two-node Intel Ice Lake setup interconnected via 100GE Intel E810CQ interfaces. The benchmarking workflow used iperf3 dynamically linked with libvcl_ldpreload.so, enabling transparent redirection of UDP socket operations to the HostStack without modifying the application code. Two traffic scenarios were evaluated: a baseline with one UDP stream achieving 15.5 Gbps, and a scaled case with one client transmitting ten concurrent streams reaching 28.1 Gbps. The results demonstrate an 81% throughput gain with minimal variance, confirming the HostStack’s efficiency under concurrent flow conditions. Integration via LD_PRELOAD proved stable, transparent, and production-compatible. These findings validate the scalability, reproducibility, and operational viability of edge-based UDP termination for HFC modernization. The proposed model offers a low-disruption upgrade path for cable operators, enabling phased deployment of user-space transport enhancements with CI/CD compatibility, improved upstream utilization, and support for multi-session, latency-sensitive applications such as telemetry, video streaming, and cloud synchronization. In addition, this architecture enables measurable performance improvements without hardware changes, costly rewiring, or disruptions to existing customer services.

KEY ASPECTS FOR THE DEVELOPMENT OF INFORMATION AND MEASURE-MENT SYSTEMS FOR DETERMINING ENVIRONMENTAL POLLUTION

Liubomyr SIKORA, Nataliia LYSA, Olga FEDEVYCH, Nazarii KHYLIAK — Thu, 26 Jun 2025 00:00:00 +0300

The intensification of production at petrochemical, construction, industrial, and energy companies and the aging of equipment at major technological and energy facilities lead to increased emissions of toxic and dusty substances into the air, soil, and water. As a result of the ecosystem cycle, these chemical compounds enter the soil and water, causing pollution.

The main task of environmental expertise is to determine the degree of risk and safety of industrial activity, organize a program of expert assessment of industrial production facilities, establish compliance of facilities with the requirements of environmental legislation, examine the quality of natural resources, form a balance of quality criteria for the environmental safety of facilities and the environment, assess the negative impact of industrial and municipal structures on the environment, and expertly evaluate programs for the introduction of new technology.

Air monitoring is necessary to detect the effects of pollutants and their impact on: corrosion of structures, erosion of land resources, impact on human health, impact on flora and the environment, water pollution, and food contamination.

The article highlights the basic concepts for building information and measuring systems (laser concentrometer and opto-galvanic sensors) for rapid analysis of environmental pollution such as air, water, and soil in emergency and extreme situations. The air and water pollution was detected around energy facilities and various industrial production facilities that are at risk of military attack. The article describes the development and construction of a laser information and measurement system for measuring dust in the atmosphere and presents the results of a study of the chemical pollution of water wells in various industries.

INFORMATION TECHNOLOGY FOR PREDICTING THE RELIABILITY LEVEL OF TEXT MESSAGES

Roman ANDRIIV, Vsevolod SENKIVSKYY — Thu, 26 Jun 2025 00:00:00 +0300

The research presents the results of the creation of an intellectualized information technology for predictive analysis of the reliability of text information messages, formed on the basis of concepts and tools of fuzzy logic. The use of the fuzzy set apparatus makes it possible to take into account the semantic ambiguity inherent in a natural language, as well as to formalize qualitative expert assessments by using linguistic variables, fuzzy term-sets and a rule base of the “if-then” type. This provides the possibility of creating adaptive decision-making models in conditions of incompleteness, inconsistency and subjectivity of input information.

The developed technology includes fuzzification of input characteristics of texts, aggregation of expert judgments, construction of a system of fuzzy rules for assessing the reliability level and defuzzification of the obtained results. A concept is implemented that allows for a predictive assessment of the veracity of data even before their potential appearance in the information space. Within the framework of the proposed approach, a structured information database is formed, which establishes a relationship between the input variables, their linguistic nature, permissible ranges of values of the universal term-set, as well as clearly defined linguistic terms used for qualitative interpretation of parameters. Based on the performed structuring of linguistic variables of the studied process, a method of logical inference is developed, which represents a multi-level hierarchy of relationships between database components and determines the algorithm for calculating the message reliability indicator. The method is based on a knowledge matrix, leading to the construction of fuzzy logical equations, which provide the calculation of normalized values of membership functions of linguistic variables at the division points of the universal set. The result is the defuzzification of the fuzzy set “the indicator of the reliability level of text information messages” and the calculation of its value using the centre of mass formula, taking into account the input data. As a result of the study, a structural model of the information technology component of assessing the veracity of news content is developed.

COMPUTERISED BLOOD PRESSURE MONITORING IN OUTPATIENT SETTINGS

CHUIKO GENNADY, Olga YAREMCHUK — Thu, 26 Jun 2025 00:00:00 +0300

The paper presents the measurement of a normal 24-hour heart rate and blood pressure analysis of an anonymous patient. The object of study in this paper is computer processing of outpatient blood pressure monitoring. The goal is to mathematically model the data as a sum of relatively smooth trends and detrended fluctuations. Tasks: decomposition of the primary series by two independent methods, stability and spectral analysis of the shifted fluctuations using the Wiener-Hinchin theorem, and proving the self-similarity of such fluctuations. The methods used are: singular spectrum analysis, exponential smoothing of the simulation and analysis of autocorrelation functions. The following results are obtained. The dataset is a sum of fairly smooth trends and detrended fluctuations; blood pressure trends have certain nighttime minima; detrended fluctuations are fractional Gaussian noise with a Hurst index of about (0.80 ± 0.016), the energy spectra of detrended fluctuations were found for the first time. Scientific novelty of the results: 1) the measured 24-hour heart rate and blood pressure analyses are decomposed into fairly smooth trends and detrended fluctuations; 2) trends allow for a more reliable assessment of 24-hour, nightly and daily average blood pressure values, which are the main indicators of a series of blood pressure measurements and monitoring; 3) detrended fluctuations contain other valuable diagnostic information, such as short-term blood pressure variability or persistence index. 4) fluctuation analysis provides information about the power spectra of the blood pressure monitoring series and their similarity to the spectra of fractional Gaussian noise; 4) knowledge of short-term changes in blood pressure is the basis for constructing informative repeatability graphs for blood pressure monitoring; 5) detrended fluctuations are identified as fractional Gaussian noise, which is a self-similar stochastic process.

WORKLOAD BALANCING IN THE TEST CASE SCHEDULING: A METHEMATICAL APPROACH

Iryna PIKH, Oleksii BILYK — Thu, 26 Jun 2025 00:00:00 +0300

Efficient scheduling of test cases is a critical task in environments where execution resources, such as testers or test environments, are limited and subject to individual availability constraints. In this paper, we propose a flexible and extensible mathematical model for optimizing test scheduling based on discrete time blocks. Each test case has a fixed duration and must be assigned to exactly one compatible tester. Testers, in turn, may be unavailable at specific time blocks due to pre-scheduled meetings or fixed breaks, such as lunch. The scheduling objective is to minimize the makespan, defined as the latest finish time among all scheduled tests. The model is formulated as a mixed-integer linear programming (MILP) problem that integrates testers' compatibility and availability constraints with task assignments into a unified framework. In contrast to models that assume testers are always available or disregard personal schedules, our method incorporates individual availability constraints for more realistic planning. The model is assessed on a synthetic scenario involving multiple testers with defined break times and varying task compatibility, and the resulting schedule is visualized with Gantt charts. The proposed formulation serves as a foundation for more advanced scheduling systems in quality assurance and resource-constrained testing workflows.

ANALYSIS OF BIOMETRIC ACCESS CONTROL SYSTEMS

Houda El BOUHISSI, Pavlo YURKO — Thu, 26 Jun 2025 00:00:00 +0300

The paper presents a method and a software-hardware tool for an access control system based on biometric data. The method involves the collection, processing, and verification of biometric features such as fingerprints, facial recognition, or iris scans to authenticate individuals. The system ensures secure access while minimizing the risks associated with traditional password-based security systems. The software-hardware tool integrates biometric sensors, data storage, and authentication algorithms to provide an efficient and secure means of controlling access to protected areas or resources. This approach aims to enhance security, streamline user access, and reduce the likelihood of unauthorized access or identity theft.

ARTIFICIAL INTELLIGENCE APPROACH TO IDENTIFYING PROPAGANDA TECHNIQUES AND OBJECTS, TAKING INTO ACCOUNT ETHICAL AND LEGAL ASPECTS

Maryna MOLCHANOVA, Pawan Kumar DUTT — Thu, 26 Jun 2025 00:00:00 +0300

The article explores the ethical and legal aspects of applying artificial intelligence (AI) technologies to detect propaganda techniques in textual content. The study presents a multi-level approach to identifying signs of propaganda in textual data, recognizing common rhetorical strategies of influence, and establishing semantic links between the detected techniques and their respective targets. The consistent use of neural network models is justified, as it ensures both classification accuracy and transparency of the obtained results through the application of local interpretability methods. The paper presents experimental results based on a corpus of Ukrainian-language news texts and informational messages from social media platforms. The proposed approach demonstrated alignment between the model's predictions and independent expert assessments, confirming its potential applicability in conditions with limited human oversight.

Special attention is given to the compliance of the proposed system with existing regulatory frameworks, including constraints on automated decision-making, the user's right to explanation, and the prevention of discriminatory effects resulting from biased training data. The study addresses risks associated with misclassification, potential impacts on freedom of expression, and the accountability of developers in cases where the system is applied in automated content moderation scenarios.

The integration of interpretability tools into neural network analysis is proposed as a core design principle to ensure adherence to ethical AI standards. Based on the obtained findings, the study concludes that the development of such systems requires the simultaneous consideration of technical effectiveness, legal compliance, and social responsibility, which are essential conditions for their safe implementation in the practice of analyzing public communications.

COMPARATIVE ANALYSIS OF COMPUTATIONAL PERFORMANCE OF MODERN PROGRAMMING LANGUAGES

Oleg ZHULKOVSKYI, Inna ZHULKOVSKA, Hlib VOKHMIANIN, Anastasiia TKACH — Thu, 26 Jun 2025 00:00:00 +0300

The study is dedicated to the comparative analysis of the computational performance of modern programming languages in the implementation of numerical methods for solving boundary value problems in mathematical physics. The central focus of the research is the Thomas algorithm – an efficient numerical method for solving systems of linear algebraic equations with a tridiagonal matrix. The research methodology is based on a unified implementation of the Thomas algorithm for each examined programming language, ensuring identical algorithmic logic. Experimental testing was conducted on systems with sizes ranging from 10⁵ to 1.5 × 10⁷ elements for programming languages including C, C++, C#, Java, JavaScript, Go, and Python, which represent different paradigms and approaches to computation. The obtained results demonstrate significant differences in the performance of various programming languages. It was established that low-level compiled languages exhibit the highest execution speed, especially for large problem sizes. In contrast, interpreted languages show significantly lower performance, which becomes more pronounced as the computational workload increases. The study experimentally confirmed the impact of compiler optimization modes on performance, revealing differences of up to 70% depending on the language and optimization level. The scientific novelty of this work lies in the comprehensive investigation of programming language performance in the context of numerical modeling by comparing their characteristics when solving mathematical problems. Future research will include an in-depth study of the impact of processor architecture, compiler optimization mechanisms, and runtime environment implementation on the performance of computational algorithms, as well as an expansion of the range of numerical methods and programming languages analyzed.

IMAGE ROTATION-INVARIANT REPRESENTATION VIA REMOVAL OF ORIENTATION FEATURES FROM THE ENCODER LATENT SPACE

Anna BEDRATIUK — Thu, 26 Jun 2025 00:00:00 +0300

In many computer vision tasks, accurate object recognition is complicated by arbitrary object orientations. Ensuring rotation invariance is critical for improving classification accuracy and reducing errors related to the varying placement of objects. This issue is particularly important in real-world environments, where object orientation is rarely controlled.

The goal of this study is to develop a method that allows separating rotational features from the semantic essence of an object, while preserving high classification accuracy after removing orientation-related components. This approach enables the construction of models that remain effective under a wide range of input perspectives, thus improving robustness in practical applications.

The proposed method is based on using a convolutional variational autoencoder trained on a dataset of images subjected to various rotation angles. Linear regression is then used to identify those latent components that correlate most strongly with the rotation parameter. These components are removed, and the remaining features are used for classification. Additionally, image reconstruction is performed from the reduced latent vector to visually validate rotation invariance and evaluate the preservation of object shape.

Experiments on a synthetically rotated binarized digit dataset show that removing “rotational” components from the latent space does not lead to a critical drop in overall classification accuracy. Instead, the removed components primarily influence orientation, supporting the possibility of clearly disentangling geometric and semantic features. Images reconstructed without these components remain recognizable but appear rotation-normalized, indicating the suppression of orientation information. A quantitative assessment confirms that the loss in accuracy is proportional to the contribution of removed components in the rotation regression.

The scientific novelty of this work lies in introducing a simple and reproducible method for removing orientation-related features from the latent space of an autoencoder without modifying the model architecture or introducing specialized regularizers. The practical significance of the method is in reducing the influence of arbitrary object orientation on recognition accuracy, thereby increasing the universality and reliability of vision systems in uncontrolled settings. The proposed approach may be useful for building classifiers capable of handling images with varying or unknown orientations during data collection.

SYSTEM FOR CYBERSECURITY EVALUATION OF CORPORATE NETWORKS

Ihor RAMSKYI, Andriy DROZD, Oleksii LYHUN — Thu, 26 Jun 2025 00:00:00 +0300

In the context of rapidly increasing cyber threats and the growing complexity of corporate IT infrastructure, ensuring a reliable and proactive approach to cybersecurity is becoming critically important for organizations of all sizes. Traditional cybersecurity assessment methods often fail to keep up with the dynamic nature of emerging threats – necessitating the development of more adaptive and intelligent evaluation systems. This article presents a comprehensive modular system for assessing the cybersecurity level of corporate networks – offering a holistic view of the security landscape by integrating both technical and organizational indicators.

The proposed system utilizes self-organizing analytical methods to dynamically process large volumes of data related to vulnerabilities, configuration states, and network behavior patterns. Through intelligent algorithms and adaptive learning, the system is capable of autonomously detecting anomalies, evaluating potential attack vectors, and correlating threats with the network’s weak points. Additionally, the inclusion of organizational factors – such as policy compliance, user behavior, and access structures – enables a more contextual and in-depth risk assessment.

A key advantage of the system is its ability to perform real-time monitoring and dynamic risk evaluation – empowering decision-makers to take informed actions in response to incidents. The system's architecture supports scalability and compatibility with existing security tools and network management platforms.

To validate its effectiveness, the system was implemented and tested in a simulated corporate environment reflecting modern structural and operational challenges. The experimental results confirmed its capability to identify vulnerabilities, prioritize responses, and enhance overall cyber resilience.

This research contributes to the advancement of next-generation cybersecurity assessment tools – ensuring the continuous improvement of corporate defense mechanisms in an ever-changing cyber landscape.

IDENTIFICATION OF SOUNDS BASED ON THE HILBERT-HUANG TRANSFORM FOR THE TASK OF DETECTING UAVs

Mariia BIERIESTOVA , Volodymyr MOROZ — Thu, 26 Jun 2025 00:00:00 +0300

This paper investigates the identification of drone acoustic signatures via the Hilbert–Huang Transform combined with machine- and deep-learning techniques. After high-pass filtering at 80 Hz, each 3-s recording is split into 25 ms frames with 50 % overlap; every frame undergoes DCT pre-whitening, HHT processing, and extraction of 13 MFCCs. The resulting averaged feature vectors—drawn from 2 075 yes_drone and 266 unknown samples—are fed into a lightweight two-layer perceptron (64 + 32 ReLU units, sigmoid output). On a held-out test set the network reaches an overall accuracy of 0.95, with a drone recall of 0.99 and an F1-score of 0.78 for the unknown class. These results outperform a baseline MFCC + SVM system and approach the performance of deeper CNN architectures, while remaining computationally suitable for real-time embedded deployment. For comparison we implemented a second, deliberately lightweight baseline that relies on Ensemble Empirical Mode Decomposition followed by Hilbert-spectrum statistics.

CALCULATION OF FOREST COVER CHANGE USING LANDSAT SATELLITE SERVICE AND R PROGRAMMING AND DATA ANALYSIS LANGUAGE

Oleksandr MELNYKOV, Viktoriia DENYSENKO — Thu, 26 Jun 2025 00:00:00 +0300

The problem of calculating the change in the level of forest cover in a selected forestry is considered. It is stated that the authors previously developed software for calculating forest cover and processing information on forest stands using the example of a village in the Kharkiv region. Calculations for comparing forest cover over several years using the Global Forest Watch resource, which marks in different colors the places where new stands are planted or existing ones are cut down, are also described. A number of features and shortcomings of this resource are identified. To improve the calculations, it is proposed to use satellite images of the Landsat / TimeSync project. Images of a separate forestry were taken from this resource for the period from 1984 to 2024. The resulting images were loaded into the updated application, then divided into parts (sections). The model previously created by the authors with a list of input factors containing indicators of the percentage of green color in the selected area and in neighboring areas for three years (the significant and the previous two) was chosen as the basis for the forecasting model. The predicted factor is the percentage of green color in the studied area. The formulated problem is proposed to be solved by the method of artificial neural networks in the environment of the programming and data analysis language R. A script was created in this language that not only builds an artificial neural network but also determines the best architecture and effective method of training a neural network for the selected data set. The calculation of the change in forest cover on the site is presented; a forecast is made for the last year, which provides an error of 2.3%. It is determined which architecture of the neural network provides the best result. The results of the calculations indicate high accuracy of the forecast.

NEURAL NETWORK DECISION SUPPORT SYSTEM FOR FORMULATING A RACING TEAM STRATEGY

Iryna GITIS, Veniamin GITIS — Thu, 26 Jun 2025 00:00:00 +0300

The research is aimed at studying the features of strategy formulation in Formula One auto racing, the factors that influence this process, and identifying ways to increase the effectiveness of the strategy through the use of artificial intelligence methods. During Formula One races, teams face a large number of challenges related to various aspects, in particular tire wear and degradation. Teams deal with them through the implementation of the strategy – determining the correct pit stop moment and choosing the appropriate type of tires. Based on the determination of factors influencing the rate of tire wear and degradation, an original feature space was formed. Analysis of the strategies of race winners in previous seasons showed that the maximum number of pit stops was three. Using the principal component analysis, a study of the original feature space was conducted, the least relevant features were identified and subsequently removed from the original data set. To solve the problem, a system was proposed and built, consisting of four modules, each of which is a multilayer feedforward artificial neural network. Using the inequality proposed by Widrow and the sample-based estimation of the Lipschitz constant, the minimum required number of neurons of the hidden layers for each neural network module was determined. During the training process, their number was specified to achieve acceptable prediction results. AdaMax was used as an optimization algorithm, and the Huber loss function was chosen to calculate the error of the networks output. The mean squared error of the resulting system prediction on the test set was 0.1. The use of such system will reduce the decision-making time of teams when formulating a racing strategy, which in turn will contribute to achieving higher results in races.

CYBER-PHYSICAL SYSTEM FOR DETERMINING SOIL PARAMETERS

Yurii VOICHUR, Illya PAYONK — Thu, 26 Jun 2025 00:00:00 +0300

The relevance of a cyber-physical system for determining soil parameters in Ukraine is determined by several important factors, including climate change, declining soil fertility, and the need to implement efficient technologies to ensure sustainable agriculture. In Ukraine, where a large part of the economy depends on the agricultural sector, accurate soil monitoring is a key aspect to increase the efficiency of agricultural production, optimize the use of water and land resources, and reduce the cost of fertilizers and pesticides.

Cyber-physical systems can provide timely data collection on soil moisture, temperature, pH, and other critical soil parameters, allowing farmers to respond quickly to changes in environmental conditions. Such systems can reduce the negative impact of excessive irrigation and optimize the use of water resources, which is especially important in the face of drought, which is increasingly common in Ukraine due to climate change.

These systems also allow for accurate forecasts of yields and soil conditions, as well as the development of individualized recommendations for each field or plot. Since Ukraine has a wide variety of climatic conditions and soil types, cyber-physical systems are able to adapt to different agricultural needs, making them extremely useful for the development of precision agriculture.

The introduction of such technologies helps not only to preserve natural resources but also to improve the economic efficiency of agriculture. Therefore, the development and implementation of cyber-physical systems for soil monitoring is an extremely important step for the sustainable development of Ukraine's agricultural sector. Therefore, our research is devoted to the development of a method and a cyber-physical system for determining soil parameters.

The cyber-physical system for determining soil parameters consists of three levels: the level of sensors, the level of the controller to which the sensors are connected, and the system for collecting, monitoring, and managing data in real time. To build a cyber-physical system for determining soil parameters, we selected sensors, selected a controller, selected a data transmission standard, and developed a method for collecting, monitoring, and controlling data. The proposed method of data acquisition, monitoring and control for the upper level of the cyber-physical soil parameterization system allows for efficient data acquisition, monitoring and control in a cyber-physical system with various parameters stored in real time.

PREDICTION MODEL FOR POTENTIAL VEHICLES COLLISION

Oleksandr BYZKROVNYI, Kyrylo SMELYAKOV — Thu, 26 Jun 2025 00:00:00 +0300

The research subject is road crash accident nature and approaches for its preventions or predictions in the real-time using computer vision algorithms and usage of edge devices. The goal of this research is to create a model for prediction of potential vehicles collision, which works for real-time. The methodology used in the research is a combination of computer vision model TrafficCamNet_1.3 output with the math approaches to determine the possible vehicles collision. The exact math methods include calculation of cars’ movement projections and usage it for checks whether vehicles collision may occur or not. The experiments setup is based on the scenarios designed using BeamNG.tech, usage of Nvidia Jetson Orin Nano as a platform for running real-time classification and determination of possible road crash accidence. The main results of this research are outlining the exact time spent for having car stopped before crash, exact cars’ characteristics and case setup and the percentage of happened road crash accidents to determine the model robustness and ability to real life introducing. As a conclusion, this research reveals the facts, that model works for the cases, when cars did not exceed allowed speed limit on the particular road. With the allowed speed, driver will be able to be notified in time and will have enough time to stop the car, otherwise amount of time to react on the threat is being significantly reduced. As a model improvement, the usage of models’ ensemble with different training dataset sizes can be considered for early car classification on the image. The results of this research can be used for building the intelligent software system for the preventions of road traffic accidence on the defined as a dangerous road parts.

REINFORCEMENT LEARNING METHOD FOR AUTONOMOUS FLIGHT PATH PLANNING OF MULTIPLE UAVS

Maksym VELYCHKO, Tetiana KYSIL — Thu, 26 Jun 2025 00:00:00 +0300

This study aims to develop a reinforcement learning method for autonomous flight path planning of multiple UAVs under real-world conditions with limited observations and multiple conflicting optimization objectives. The research proposes a multi-agent reinforcement learning approach based on Proximal Policy Optimization (PPO) combined with centralized training and decentralized execution (CTDE). Additionally, a recurrent neural network (RNN) layer is integrated into the critic and actor networks to address partial observability. The reward function is designed to balance time efficiency, safety, and area coverage. Experimental results demonstrate that the proposed method significantly outperforms independent learning approaches in terms of reward accumulation, convergence speed, and decision stability. The CTDE architecture with RNN-enhanced critics proved effective in handling the challenges of multi-agent coordination and partial observability. The trained model enables real-time trajectory planning in three-dimensional environments, surpassing traditional optimization methods. The novelty lies in the application of a multi-agent PPO architecture enhanced by RNNs under CTDE for solving real-time multi-objective optimization problems in UAV path planning. A customized reward structure was developed to simultaneously optimize safety, time, and coverage objectives without retraining. The developed method enables efficient and reliable online trajectory planning for UAV groups, making it applicable in surveillance, search and rescue, and exploration missions where rapid and adaptive decision-making is essential.

ADAPTIVE VIDEO ENHANCEMENT BASED ON BLIND DEGRADATION ESTIMATION

Mykola MAKSYMIV, Taras RAK — Thu, 26 Jun 2025 00:00:00 +0300

Video enhancement aims to restore high-quality video from degraded inputs affected by noise, blur, compression artifacts, or resolution loss. Most existing models assume a fixed degradation during training, limiting their robustness to real-world scenarios with unknown and varying distortions. In this paper, we propose a quality-aware video enhancement framework that explicitly estimates the input degradation level and conditions the restoration process accordingly.

Our method consists of a lightweight degradation level estimation module that predicts a quality score for each frame, and a conditional enhancement network that dynamically adjusts restoration strength based on the estimated degradation. Unlike static models trained for a single degradation type, our system adapts to diverse distortions, applying appropriate enhancement strategies for different quality levels.

Extensive experiments on standard datasets such as Vimeo-90K and REDS demonstrate that our method consistently outperforms strong baselines, including BasicVSR, EDVR, and others, particularly under blind degradations. The proposed framework improves PSNR, SSIM, and LPIPS scores, while maintaining temporal consistency and introducing only minor computational overhead. These results highlight the potential of explicit quality estimation for achieving robust and perceptually faithful video restoration across varying real-world conditions.

PROCESS MODEL FOR ENSURING THE SECURE FUNCTIONING OF INTERNET OF THINGS DEVICES BASED ON A HEURISTIC SEARCH ALGORITHM

Miroslav KVASSAY, Oleh BONDARUK, Vadym DIDUKH, Olha ATAMANIUK — Thu, 26 Jun 2025 00:00:00 +0300

The proliferation of Internet of Things (IoT) devices in modern critical infrastructures has brought new challenges related to their secure functioning. Traditional cybersecurity mechanisms such as firewalls, antivirus software, and intrusion detection/prevention systems are often ineffective in IoT environments due to device heterogeneity, limited computing capabilities, decentralized control, and physical vulnerability of nodes. To address these challenges, the paper proposes a process model for ensuring the secure functioning of IoT devices, utilizing a heuristic search algorithm to optimize device deployment with minimal security risk. The proposed model is structured as a multi-stage data processing pipeline that encompasses the full decision-making lifecycle: from gathering network data and identifying vulnerabilities, to generating attack graphs, simulating deployment scenarios, assessing risk, and selecting the optimal deployment strategy. The core of the model is a heuristic-based optimization mechanism (DFBnB – Depth-First Branch and Bound), which efficiently searches a large decision space structured as a binary tree of deployment options. Each deployment scenario dynamically modifies the attack graph, allowing the model to evaluate security risks in real time based on parameters such as the number and length of attack paths, the presence of vulnerabilities, and the privilege escalation potential. Two optimization goals are considered: full deployment of all IoT devices with minimal risk, and maximization of deployed devices without increasing existing risk indicators. The model formalizes these goals using objective functions and integrates real-time heuristics for effective pruning of suboptimal solutions. Experimental validation was conducted using a simulated organizational network with the set of hosts and IoT devices, under various placement scenarios. The results demonstrated that the heuristic approach significantly reduces computation time compared to full search, while maintaining a high level of network security. The optimized deployments preserved core network resilience and enabled safe integration of devices without increasing security risks. Overall, this research offers a scalable and adaptable framework for secure IoT deployment, which can serve as the foundation for intelligent, risk-aware security management in dynamic and heterogeneous network environments.