| Autor: |
S.A. Yamashkin, A.A. Kamaeva |
| Jazyk: |
English<br />Russian |
| Rok vydání: |
2021 |
| Předmět: |
|
| Zdroj: |
Известия высших учебных заведений. Поволжский регион:Технические науки, Iss 3 (2021) |
| Druh dokumentu: |
article |
| ISSN: |
2072-3059 |
| DOI: |
10.21685/2072-3059-2021-3-2 |
| Popis: |
Background. In the modern world, component development methods are becoming increasingly popular. They allow not only to quickly solve the assigned tasks, but also to meet the high requirements for the performance and reliability of the created software products. The use of the block approach in machine learning is a breakthrough method that makes the development of complex neural network architectures several times easier. The purpose of this work is to study the methods of component development and the technology of graph-symbolic programming for solving the problem of visual programming of neural networks. Materials and methods. The work develops methods and algorithms for configuring neural network models based on software systems that implement a graphical interface for creating neural networks using graph-symbolic programming based on the JavaScript programming language. Results. Comprehensive research has been carried out in the field of visual programming of neural networks using component development methods. A graphical interface has been developed for configuring neural network architectures. Conclusions. The presented approach to visual programming of neural networks simplifies the development process, avoids errors and creates more efficient and reliable systems. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
