Publication: The Impact of Input Data Density on the Performance of Graphic Neural Networks
Дата
2024
Авторы
Bondareva, N. A.
Journal Title
Journal ISSN
Volume Title
Издатель
НИЯУ МИФИ
Аннотация
The paper provides a brief overview of generative neural networks and considers the role of information in training generative neural networks. In the digital environment, each object is surrounded by a vast information field, including unordered information and a set of references to it. The density of the object's information field determines the ability of technologies such as artificial intelligence to recreate its image based on the collected data. The more data is available, the more accurately and completely the digital image can be recreated. The paper considers a number of problems arising from the use of text-to-image networks and possible methods for solving them. The article considers various aspects of the role of personal data and possible ethical and social consequences in the era of generative technologies, as well as the prospects and risks of further development of generative neural networks in specialized areas such as medicine and manufacturing. The rapid development of neural network technologies can have a significant impact on education and social phenomena.
Описание
Ключевые слова
Text-to-image , Information field density , Computer graphics , Neural network , Computer vision and pattern recognition , Machine learning
Цитирование
N.A. Bondareva. The Impact of Input Data Density on the Performance of Graphic Neural Networks (2024). Scientific Visualization 16.5: 109 - 119, DOI: 10.26583/sv.16.5.08