Перегляд за автором "Turuta, O."

Зараз показано 1 - 5 з 5
  • Зображення мініатюри
    Публікація
    A survey of methods of text-to-image translation
    (ХНУРЕ, 2019) Konarieva, I.; Pydorenko, D.; Turuta, O.
    The given work considers the existing methods of text compression (finding keywords or creating summary) using RAKE, Lex Rank, Luhn, LSA, Text Rank algorithms; image generation; text-to-image and image-to-image translation including GANs (generative adversarial networks). Different types of GANs were described such as StyleGAN, GauGAN, Pix2Pix, CycleGAN, BigGAN, AttnGAN. This work aims to show ways to create illustrations for the text. First, key information should be obtained from the text. Second, this key information should be transformed into images. There were proposed several ways to transform keywords to images: generating images or selecting them from a dataset with further transforming like generating new images based on selected ow combining selected images e.g. with applying style from one image to another. Based on results, possibilities for further improving the quality of image generation were also planned: combining image generation with selecting images from a dataset, limiting topics of image generation.
  • Зображення мініатюри
    Публікація
    Evolving flexible neuro-fuzzy system for medical diagnostic tasks
    (IJCSMC, 2015) Turuta, O.; Perova, I.; Deineko, A. O.
    In the paper architecture and training method for evolving flexible diagnostic neuro-fuzzy-system are investigated. The proposed system is simple in numeric realization and characterized by a high learning rate and flexibility, that make possible to use it in conditions of small training sets and on big data sets, coming to processing in online-mode.
  • Зображення мініатюри
    Публікація
    Evolving Neural Network for Kernel Principal Component Analysis
    (IJCSMC, 2015) Turuta, O.; Deineko, A. O.; Perova, I.; Kutsenko, Y.; Shalamov, M.
    In the paper kernel evolving neural network and its learning algorithm are investigated. The proposed system solves the problem of finding the eigenvectors and the corresponding principal components in on-line mode in an environment where hidden in the experimental data interdependencies are nonlinear and can change throw time.
  • Зображення мініатюри
    Публікація
    Geoscience Laser Altimeter System sparse ICESat data processing based on F-transform
    (2019) Yerokhin, A.; Babii, A.; Turuta, O.
    In the paper, we present the 3D filtering method of surface approximation based on FT-smoothing proposed to estimate glacier thickness change. Our method based on ICESat/GLAS Altimetry Data and SRTM-DEM Data. Processed data of satellite measurements at the coordinate surface usually placed sparsely, collected in “Satellite track”. For interpolation, we are using 2D membership function. This approach is kind of a generalization of the single-dimension method. The result provides estimated by RMSE, we have managed a number of fuzzy components for improving RMSE.
  • Зображення мініатюри
    Публікація
    Information model for heat and mass transfer processes evaluation
    (2019) Yerokhin, A.; Zatserklyanyi, H.; Babii, A.; Turuta, O.; Zolotukhin, O.
    Розглянуто проблему вибору моделі і методу оцінювання теплопровідності, конвективного і радіаційного теплообміну. Обрано структурну схему взаємопов'язаного та взаємозалежного процесу тепломасопереносу. Реалізовано і апробовано інформаційну модель для оцінювання взаємопов'язаного процесу тепломасопереносу і реалізації його програми. The problem of choosing the model and method of estimation of thermal conductivity, convective and radiation heat exchange is considered. The structural scheme of the interconnected and interdependent process of heat and mass transfer is selected. An information model has been implemented and tested to evaluate the interconnected process of heat and mass transfer and its program implementation.