Перегляд за Автор "Tymofiiev, I."

Зараз показуємо 1 - 3 з 3
  • Ескіз
    Документ
    Approach for Using Neural Network BERT-GPT2 Dual Transformer Architecture for Detecting Persons Depressive State
    (2024) Mazurets, O.; Tymofiiev, I.; Dydo, R.; Мазурець, Олександр Вікторович
    The paper proposed the method of using neural network BERT-GPT2 dual transformer architecture for detecting persons depressive state designed to transform input data in the form of text and trained neural network BERT-GPT2 dual transformer architecture model into output data in the form of the numerical assessment of the presence of persons depressive state. Experiments were conducted with the use of the given developed software complex for detecting persons depressive state, which testify to the correctness of the proposed approach. From the performed performance study, the dual architecture did not make a single error during classification
  • Ескіз
    Документ
    Formalization of Fabric Recycling Methods Depending on Raw Material Composition for Intelligent Decision Support Systems
    (2026) Zagorodnya, A.; Tymofiiev, I.; Molchanova, M.; Mazurets, O.; Мазурець, Олександр Вікторович
    The paper proposes a formalization of fabric recycling methods depending on raw material composition for intelligent decision support systems in textile recycling. The study focuses on combining computer vision, machine learning, and rule-based decision mechanisms to support automated textile waste sorting and processing. The proposed approach includes classification of fabrics into raw material categories, assessment of prediction reliability, and generation of recommendations for further handling, such as reuse, mechanical recycling, chemical recycling, downcycling, or manual inspection. The methodology emphasizes not only fabric recognition but also the intelligent selection of the most appropriate recycling scenario. Experimental results demonstrate that computer vision models can achieve high accuracy in distinguishing natural and synthetic fabrics, creating a practical basis for automated textile recycling systems. The proposed approach can be applied in textile sorting lines, recycling centers, and digital decision support services for sustainable waste management.
  • Ескіз
    Документ
    Neural Network Dual Architecture for Depression Detection Using Cloud Services
    (2024) Tymofiiev, I.; Mazurets, O.; Hardysh, D.; Molchanova, M.; Мазурець, Олександр Вікторович
    A method of detecting a depressive state by means of NLP was developed, which is designed to transform input data in the form of text and a trained neural network model of dual architecture into output data in the form of a numerical assessment of the presence of a depressive state. The proposed method differs from analogs in that it combines a two-stream architecture, which is based on the use of two parallel neural networks, each of which specializes in the analysis of different aspects of the text - syntactic and semantic.