Бионика интеллекта

Постійний URI для цієї колекції

https://openarchive.nure.ua/handle/document/11

Перегляд

Зараз показано 1 - 20 з 345

3D технологии в производстве ювелирных изделий
(ХНУРЭ, 2016) Вовк, А. В.; Кузнецова, В. С.
Исследованы современные технологии производства ювелирных изделий от эскиза до готового изделия. Проведён анализ существующих программных продуктов, позволяющих проводить моделирование ювелирных изделий с дальнейшим выводом на ювелирный 3D принтер. Предложен алгоритм моделирования, оптимизирующий построение 3D модели. Определена наиболее оптимальная технология 3D печати.
A neural network approach for the auto matic selection of a complex of rehabilitation exercises
(ХНУРЭ, 2021) Butsenko, M. O.; Afanasieva, I. V.; Golian, N. V.; Kameniuk, N.
This article is devoted to solving the problem of automatic selection of a set of rehabilitation exercises during injuries, considering the state of the human cardiovascular system through the use of neural networks. To solve this problem, it was necessary to choose one of two classical approaches – multiclass classification or multilabel classification, each of which solves the problem of data classification through its own algorithm, and use the selected neural network architecture to create a software system. While working on this system, it was also necessary to solve certain problems related to each of these approaches (the need for a large sample due to the large number of exercises that the system should recommend) or a specific approach (inability to select multiple exercises at once – for Multiclass Classification, lower productivity and the number of supported programming languages – for Multilabel Classification). Samples of different sizes (from 1 million records and more) were used to train the neural network, which were generated through a self-written program that generated a given number of records and wrote them to a .CSV (commaseparated values) file.
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.
Analysis of Net Work Performance Under Selfsimilar System Loading by Computer Simulation
(ХНУРЭ, 2008) Kirichenko, L. O.; Radivilova, T.
The simulation have shown that management of selfsimilar traffic allows to improve quality of network service and avoid overflow of the buffer memory.
Applying gradient boosting as a stac king algorithm over bottleneck features to achieve high image classification accuracy
(ХНУРЭ, 2021) Golian, N.; Afanasieva, I.; Golian, V.; Panchenko, D.
With the development of the Internet, making many images available online for analysis, object recognition software is gaining more and more attention from researchers. Factors are driving the development of computer vision today: mobile devices with built-in cameras, the availability of computing power, the availability of computer vision and analysis equipment, and new algorithms such as convolutional neural networks that take advantage of the power of hardware and software. The work is generally devoted to the consideration of the problem of image classification using convolutional neural networks. And in particular, one of the most popular and applied in practice machine learning algorithms − gradient boosting applied to the bottlenecks of deep convolutional neural networks. It also discusses three scenarios for applying gradient boosting to bottlenecks extracted from the last convolutional layer of the neural network. The essence of boosting, as well as of other ensembles of algorithms, is to collect one strong from several weak models. The general idea of boosting algorithms is to consistently apply predictors so that each subsequent model minimizes the error of the previous one. Gradient boosting works by sequentially adding new models to past models so that errors made by previous predictors are corrected.
Detection of blood cells
(ХНУРЕ, 2019) Chetverykov, G. G.; Tereshchenko, G.; Konarieva, I.
The structure of the medical image analysis system is considered. The algorithm of the blood cell recognition system is given. Formulated the main tasks to be solved during the morphological analysis of blood. The requirements for the algorithm in determining the leukocyte formula and the detection of blood corpuscles on a smear were determined. A model of color-brightness characteristics is proposed for describing typical images of a blood smear. The threshold values of the sizes of objects are determined when searching for cells. A histogram of the brightness of a typical field of view was investigated. A two-step algorithm for detecting blood cells is described, as well as an algorithm for constructing a dividing line on the plane of relative colors. The results of experiments on real preparations are given. The causes of detection errors are considered.
K-means approach in tumors cell color segmentation in lab color space
(2016) Potapov, S. N.; Kulishova, N. Ye.
Digital microscope images are becoming increasingly important in the diagnosis of serious diseases such as cancer. The observations are carried out on the immunohistochemical preparations that change color under the influence of specific markers. Automatic selection of the marked areas and their analysis allows identifying the disease in its early stages. For the detection of marked areas a two-step segmentation method using k-means is proposed: on the first stage a choice of all possible marked image areas is carried out, and on the second - their partition according to the marker expression level. Segmentation is performed in the color space Lab, which allows compensating colors differences caused by marking chemical reactions variations. To evaluate the expression level using - lightness. The marker expression level is evaluated by one of the color space coordinates - lightness L.
Neural network approach for emotional recognition in text
(ХНУРЕ, 2019) Nazarenko, D. S.; Afanasieva, I. V.; Golian, N. V.
The article is devoted to one of the most popular trends in the field of IT today – natural language processing, in particular, the extraction of emotions from the text using the neural network approach. The main task was to solve the problem of the high costs of time and human resources for companies to receive feedback from users and process emotional reactions of the second one. That to decide the task it was necessary to make modelling and learn neural network using own architecture based on the backpropagation algorithm that to recognize the emotional component in the text.The emotional component of reviews was used as a metric for evaluating user reactions. It was decided to work with five types of emotions that will help to provide better results. The neural network architecture consists of interconnected layers: embedding, bidirectional LSTM, pooling, dropout layers and two dense layers. For the neural network learning was selected an open dataset consisted of 47,288-tagged posts from Twitter. As a result, the F-measure on the test dataset was 0.62 and which is a worthy indicator in comparison with large business solutionsю
Principles of searching and sorting optimization in social networks using a multi-factor assessment system
(ХНУРЕ, 2019) Shopynskyi, M. V.; Golian, N. V.; Afanasieva, I. V.
The analysis of social networks, which focuses on the relationship between social entities today is an area of active research. It is a set of tools for research, in particular, in combination with artificial intelligence methods such as machine learning, deep learning. The paper examined the current quality of the assessment of information in social networks, analyzed the methods of searching and sorting information in various social networks, as well as the process of providing recommendations to users. Social media data is an inexhaustible source of research and business opportunities. In general, social media data is information gathered from social networks that shows how users interact with content. Methods of improving search results for personalizing recommendations in social networks are given. These indicators and statistics provide an effective understanding of the strategy of behavior in social networks. The advantages and disadvantages of a multifactor assessment system are considered. The possible ways of integrating the combined system of evaluating information elements by the user to optimize search queries and filtering big data are identified.
Security in decentralized databases
(ХНУРЕ, 2019) Nazarov, A.; Kozel, N.; Gruzdo, I.; Kyrychenko, I.
Blockchain is a distributed network that records digital transactions on a publicly accessible ledger. This paper explores whether blockchain technology is a suitable platform for the preservation of digital signatures and public/private key pairs. Conventional infrastructures use digital certificates, issued by certification authorities, to declare the authentication of key pairs and digital signatures. This paper suggests that the blockchain’s hash functions offer a better strategy for signature preservation than digital certificates. Compared to digital certificates, hashing provides better privacy and security. It is a form of authentication that does not require trust in a third-party authority, and the distributed nature of the blockchain network removes the problem of a single point of failure.
Stabilization of key frame descriptions with higher order Voronoi diagram
(ХНУРЭ, 2013) Mashtalir, S. V.; Mikhnova, O. D.
Video summary is one of currently developing areas of video mining. Static summary is composed of key frames extracted from video, which fully depict its content. While extracting key frames with the help of Voronoi tessellation comparison, it has been proposed to detail frame content with higher order Voronoi diagrams. This step has lead to simplification of computational procedure compared with increasing the number of initial generator points. Key frames extracted with Voronoi diagrams have been checked for precision and recall and compared with three existing extraction techniques based on optical flow, cluster analysis and curve simplification. У статті розглядається актуальний напрямок розпізнавання відео з урахуванням вмісту. Реферування відео шляхом вилучення значимий статичних зображень, що відображають суть всього матеріалу, є темою досліджень. Автори зробили спробу реалізувати процедуру пошу- ку ключових кадрів за допомогою діаграм Вороного. Діаграми Вороного більш високих порядків пропонується використовувати при деталізації вмісту відеокадрів.
The modern quantum computing tools invstigation
(ХНУРЕ, 2020) Bozhko, I.; Chetverykov, G. G.; Karataiev, O.
The modern quantum computing tools invstigation The current paper covers the investigation of the current state of the existing tools for quantum programming including QCL (Quantum Computation Language), quantum pseudocode, Q# programming language and Quipper. Since quantum computing is one of the main research areas today, the respective tools are being created quite often. They are aimed on simplifying the development of quantum programs, on the one hand, and provide some platform for testing and running them, on the other hand. So, the authors investigated the currently available tools and provided the results in the article.
The usage and implementation of parallelism in go programming language based on the mpi interface as a message exc hange method
(ХНУРЭ, 2021) Kyrychenko, I. V.; Kolesnyk, V. V.; Shmelov, O. B.
The development of the methods for optimizing computer processes by the means of Go programming language. The resources for MPI computations were analyzed from the side of Go programming language. Proposed attempts to fabricate the ties the Go form devices hit their restriction of adaptability quick. Among the advantages of using Go programming language for implementation MPI algorithms, could be saud that it eliminates the need for the developer to manage memory and resources used by software manually, own binaries, fast and efficient compilation. Athough Golang uses several resources to create parallel computations, MPI algorithms implemented by Golang methods and techniques do not fully integrate exchange and computation. Were compared two Jacobi methods for solving partial differential equation. The results showed that Go cannot coordinate the execution of C, although Go scales a part more pleasant when using non-blocking communication when comparing the blocking C usage with the blocking Go execution and and comparing the non-blocking implementations with each other. Go programming language is used for developing massive systems that can speed up software code several times by properly converting sequential algorithms to competing ones, nevertheless MPI developers are not recommended to use it due to its complexity for implementation. As a result, there is currently almost no MPI implemented by Golang methods and techniques that would fully integrate exchange and computation
Автоматическая классификация данных на основе модели искусственной иммунной сети
(ХНУРЭ, 2014) Кораблев, Н. М.; Фомичев, А. А.
В статье рассматривается решение задачи автоматической классификации данных на основе модели искусственной иммунной сети. Для повышения скорости иммунного обучения используется пропорциональная мутация на основе аффинности дальнего предка, и конкурентно-целевой отбор клонов на этапе редактирования сети. При выделении кластеров для объектов, которые не могут быть отнесены ни к одному из исходных классов, используются стимулирующие антитела, которые становятся центрами формируемых кластеров.
Агентно-ориентированный подход на основе искусственных иммунных систем для решения задачи коммивояжера
(ХНУРЭ, 2012) Кораблев, Н. М.; Иващенко, Г. С.; Кушнарев, М. В.
В представленной работе рассматриваются особенности применения агентно-ориентированного подхода, основанного на использовании искусственных иммунных систем, для решения задачи коммивояжера. Представлены результаты экспериментальных исследований, демонстрирующие преиму- щество предложенного подхода по сравнению с применением генетических алгоритмов для решения поставленной задачи.
Адаптивне тестування знань методами логічних мереж
(ХНУРЕ, 2020) Шубін, І. Ю.; Четвериков, Г. Г.; Ляшик, В. А.; Шанідзе, Н. О.
Під адаптивним тестовим контролем розуміють комп’ютеризовану систему науково обґрунтованої перевірки й оцінювання результатів навчання, що має високу ефективність за рахунок оптимізації процедур генерації, пред’явлення й оцінки результатів виконання адаптивних тестів, що заснована на методах побудови та оптимізації логічних мереж. Алгоритми підбору й пред’явлення завдань будуються за принципом зворотного зв’язку, коли при правильній відповіді суб’єкта навчання чергове завдання вибирається більш важким, а неві- рна відповідь спричиняє пред’явлення наступного більш легкого завдання, ніж те, на яке суб’єктом навчання була дана невірна відповідь. Також є можливість завдання додаткових питань по темах, які суб’єкт навчання знає не дуже добре для більш точного з’ясування рівня знань у даних областях. Вибір алгоритмів тестування наразі фактично обмежений формами представлення тестових завдань і алгоритмами оцінювання результатів тестування. Досягнення більш високих результатів і підвищення мотивації навчання в остаточному підсумку є основною метою тестування знань. Для визначення базового алгоритму, необхідно навести сценарій роботи системи. У його основі лежить модель приймання іспиту викладачем у студента, як модель адаптивного тесту- вання. Такий вибір сценарію роботи системи обумовлений тим, що, по-перше, дана процедура історично добре формалізована, по-друге, при проектуванні тестів, їх розробнику необхідно спиратися на загальноприйняті, відомі й використовувані їм методи з мінімальною модифікацією.
Адаптивное параллельное обучение модифицированной самоорганизующейся карты Кохонена
(ХНУРЭ, 2012) Дяченко, В. А.; Михаль, О. Ф.
Рассмотрены и сопоставлены по параметрам классический и модифицированный варианты реализации обучения самоорганизующихся карт Кохонена. Проведен сравнительный анализ скорости обучения при различных типах входных данных. Полученные результаты интересны в аспекте выработки рекомендаций по распараллеливанию отдельных фаз обучения при реализации модифицированного варианта карты Кохонена на вычислительной системе, поддерживающей распределённую обработку данных.
Адаптивные нечеткие модели идентификации нелинейных объектов на основе искусственных иммунных систем
(ХНУРЭ, 2008) Кораблев, Н. М.; Сорокина, И. В.
В данной работе рассмотрены вопросы адаптации нечетких моделей идентификации нелинейных объектов, представленных в виде системы нечеткого вывода и нечеткой нейронной сети, с использованием искусственных иммунных систем. Процесс адаптации состоит в настройке формы и параметров функций принадлежности, а также параметров и структуры (в случае системы нечеткого вывода) базы нечетких правил. Показана эффективность предложенных иммунных алгоритмов адаптации моделей нечеткого вывода на тестовых функций.
Аксиоматическое введение метрики в субъективных пространствах
(ХНУРЭ, 2006) Герасин, С. Н.; Шляхов, В. В.
В статье введено понятие метрического предиката, при помощи которого можно осуществлять шкалирование субъективных образов различных сенсорных систем.
Алгебро-логические средства для идентификации и визуализации нештатных ситуаций в системах с канальной структурой
(ХНУРЭ, 2005) Ерохин, А. Л.
В статье рассматривается задача разработки алгебро-логических средств для идентификации и визуализации систем, в которых можно выделить множество уровней иерархии, таким как инженерные, электрические и информационные сети.

Перегляд

Перегляд Бионика интеллекта за назвою

Результатів на сторінку

Варіанти сортування