Перегляд за Автор "Zalutska, O.O."

Зараз показуємо 1 - 5 з 5
  • Ескіз
    Документ
    Effectiveness Research of Method for Values Forecasting of Epidemiological Danger Indicators by Means of Neural Network Modeling
    (2024) Mazurets, O.V.; Ovcharuk, O.M.; Tyschenko, O.O.; Zalutska, O.O.; Мазурець, Олександр Вікторович
    The aim of the study is effectiveness research of method for values forecasting of epidemiological danger indicators by means of neural network modeling. The method for values forecasting of epidemiological danger indicators using neural network modeling was studied, which allows, based on input data in the form of a sample of time-dependent values of a specified parameter during the studied period, to receive output data in the form of a sample with predicted values of the parameter for further forecasting of the level of epidemiological danger using neural network modelling, and uses a recurrent temporal neural network with one convolutional layer to predict parameter values from their time series.
  • Ескіз
    Документ
    Effectiveness research of using ViT neural network architecture for classifying the destroyed buildings remains
    (2025) Hladun, O.V.; Molchanova, M.O.; Zalutska, O.O.; Mazurets, O.V.; Мазурець, Олександр Вікторович
    This study explores the effectiveness of the Vision Transformer (ViT) neural network for classifying remains of destroyed buildings in post-disaster environments. A software system was developed to preprocess images, train ViT and MobileNetV3 models, and integrate them into a user-friendly application. The models, trained on real-world construction debris images from robotic systems, showed high classification accuracy. Results confirm the ViT model’s potential for reliable, automated damage assessment, supporting faster and safer disaster response.
  • Ескіз
    Документ
    Method for Analyzing the Ukrainian Language Texts Sentiment Using Natural Language Processing
    (2025) Zalutska, O.O.
    The paper focuses on intelligent sentiment analysis of text related to named entities. The proposed method combines a neural network-based natural language processing model, a lexical NLP library, and a Ukrainian sentiment dictionary. It provides results in the form of sentiment scores for named entities at the sentence and text levels, as well as an overall sentiment evaluation of the analyzed content. The relevance of the research is determined by the growing need for accurate sentiment analysis in the context of large-scale digital information flows. Identifying emotional attitudes toward specific persons, organisations, or events has essential applications in monitoring public opinion, brand perception, political discourse, and financial market analysis. The scientific novelty lies in developing and implementing a method that supports Ukrainian-language texts and evaluates sentiment across negativity, neutrality, positivity, and emotionality dimensions. The practical significance is creating a software system capable of semantic sentiment analysis of textual content, achieving higher effectiveness than translation-based approaches. The developed method can analyze public opinion, social media reactions, market trends, and individual texts.
  • Ескіз
    Документ
    Neural network classification of textiles by fiber features using microscopic images
    (2025) Zalutska, O.O.
    This study presents a reproducible neural-network method for binary textile classification based on microscopic images in the visible spectrum, aimed at distinguishing natural and synthetic fibers for circular economy applications. An open corpus of 3,107 microscope images and a unified training protocol enable fair comparison of modern architectures (ViT-B/16, EfficientNet-B0, ConvNeXt-Tiny) and ensure stable validation accuracy under realistic shooting variations, fabric deformations, and local artifacts. The approach demonstrates high classification quality on laptop-level hardware and supports practical implementation in textile sorting, laboratory composition confirmation, and quality control. Openness of data and transparency of procedures facilitate technology transfer and industrial validation, contributing to reduced waste, improved purity of secondary fractions, and lower resource consumption in textile recycling chains. Future work will focus on corpus expansion, multi-scale feature modeling, and enhanced benchmarking protocols to increase stability in complex scenes and scale the solution toward full industrial sorting systems.
  • Ескіз
    Документ
    Practical Implementation of Neural Network Method for Stress Features Detection by Social Internet Networks Posts
    (2024) Mazurets, O.V.; Sobko, O.V.; Molchanova, M.O.; Zalutska, O.O.; Yurchak, A.V.; Мазурець, Олександр Вікторович
    The article considers a neural network method for stress features detection by social internet network posts, designed for automated analysis of text messages posted on social networks in order to identify signs of stress in posts. Based on the designed functional and design architectures of the information system for detecting stress in posts, the software implementation was carried out to study the effectiveness of the developed neural network method for stress features detection by social internet network posts. The practical implementation of the neural network method has determined that the developed method allows detecting stress features in social Internet network posts with an accuracy of 90%.