2023, issue 4, p. 53-61
Received 05.09.2023; Revised 19.09.2023; Accepted 28.11.2023
Published 04.12.2023; First Online 05.12.2023
https://doi.org/10.34229/2707-451X.23.4.7
Previous | FULL TEXT (in Ukrainian) | Next
Using Deep Learning Methods for Image Generation
Violeta Tretynyk * 
, Evgeny Budzinskyi
The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"
* Correspondence: This email address is being protected from spambots. You need JavaScript enabled to view it.
Introduction. Artists have always used different mediums to express their creativity and explore their imagination. With the advancement of digital technology, artists now have access to a vast array of tools that allow them to create works of art that are more sophisticated, complex, and visually engaging than ever before. Recently, there has been a growing interest in using artificial intelligence to create images for artistic purposes.
Art imaging involves the application of algorithms and machine learning (ML) techniques to create digital works of art that can mimic the styles, techniques, and aesthetics of traditional art forms. Artificial intelligence systems can learn from massive amounts of data to create images that are incredibly realistic and detailed, as well as unique and original. The purpose of the paper. In this paper, an approach based on fuzzy logic was used to estimate the cost of housing in Kyiv. Fuzzy methods allow to apply a linguistic description of complex processes, to establish fuzzy relationships between concepts, to predict the behavior of the system, to create a set of alternative actions, to formally describe fuzzy decision-making rules.
The purpose of the article. In this work, an approach based on generative models was used for image generation. Machine learning methods, namely deep neural networks, open wide opportunities for solving the given problem.
Results. This paper considers the application of deep learning methods for image generation. A comparative analysis of existing means of image generation was carried out. A proposed modification of the generative model. The developed system generates an image of a fixed size (64x64, 256x256, 1024x1024) of an artistic nature, the minimum value of the FID index during training is 128. The program implementation of the model was performed in the Python programming language.
Keywords: convolutional neural networks, machine learning, art generation, generative competitive network.
Cite as: Tretynyk V., Budzinskyi E. Using Deep Learning Methods for Image Generation. Cybernetics and Computer Technologies. 2023. 4. P. 53–61. (in Ukrainian) https://doi.org/10.34229/2707-451X.23.4.7
References
1. Trach Y., 2021. Artificial Intelligence As A Tool For Creating And Analysing Works Of Art. Culture and Arts in the Modern World. 2021. 22. P. 164–173. https://doi.org/10.31866/2410-1915.22.2021.235907
2. Mehlig B. Machine learning with neural networks. Department of Physics University of Gothenburg Göteborg, Sweden. 2021. 241 p. https://arxiv.org/abs/1901.05639 (accessed: 03.09.2023)
3. Goodfellow I. NIPS 2016 tutorial: Generative Adversarial Networks. 2017. https://arxiv.org/abs/1701.00160 (accessed: 03.09.2023)
4. Brownlee J. A gentle introduction to generative adversarial networks (Gans). MachineLearningMastery.com. 2022. https://machinelearningmastery.com/what-are-generative-adversarial-networks-gans/ (accessed: 03.09.2023)
5. Radford A., Metz L., Chintala S. Unsupervised representation learning with deep convolutional generative Adversarial Networks. 2016. https://arxiv.org/abs/1511.06434 (accessed: 03.09.2023)
6. Liao P., Li X., Liu X., Keutzer K. The ARTBENCH dataset: Benchmarking generative models with artworks. 2022. https://arxiv.org/abs/2206.11404 (accessed: 03.09.2023)
ISSN 2707-451X (Online)
ISSN 2707-4501 (Print)
Previous | FULL TEXT (in Ukrainian) | Next