Expanding the capabilities of image generation systems by using neural networks
Keywords:
image generation, neural networks, neural network architecture, image quality problem, computational optimization problem, computational power limitationAbstract
The article provides an overview of the main challenges faced by researchers and practitioners in generating images using neural networks. Key aspects such as computational complexity, power consumption and quality of the generated images are discussed. The paper also suggests potential solutions to these problems, including optimization of neural network architectures, application of optimization techniques, and use of specialized hardware gas pedals. The prospects for research in this area are summarized, and directions for future research and innovation are outlined.
References
Зори, С. А. Реалистичная визуализация трехмерных объектов и сцен с использованием технологий объемного отображения / С. А. Зори, Е. А. Башков // Известия ЮФУ. Технические науки. Тематический выпуск «Компьютерные и информационные технологии в науке, инженерии и управлении». - 2012. - № 5(130). - С. 133-137.
Зори, С. А. Использование средств аппаратной поддержки для повышения производительности систем 3D-пространственной визуализации / С. А. Зори, А. Я. Аноприенко, Р. В. Мальчева, О. А. Авксен-тьева // Информатика и кибернетика. - Донецк: ДонНТУ, 2019. - № 1 (15). - С. 5-12.
Хайкин С. Нейронные сети: полный курс / С. Хайкин – 2-е издание. Пер. с англ. – М.: Издательский дом «Вильямс», 2006. – 1104 с.: ил.
Goodfellow, Ian, et al. Generative adversarial nets // Advances in neural information processing systems, 2014.
Zhang, R. The unreasonable effectiveness of deep features as a perceptual metric / R. Zhang // Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2018.
Постолит, А. В. Основы искусственного интеллекта в примерах на Python. Самоучитель / А. В. Постолит. – 2-е изд., перераб. и доп.– СПб.: БЧВ-Петербург, 2023 – 448 с.: ил.
Srivastava, Nitish, et al. Dropout: a simple way to prevent neural networks from overfitting // The Journal of Machine Learning Research 15.1, 2014. Р. 1929-1958.
Shorten, Connor, and Taghi M. Khoshgoftaar. A survey on image data augmentation for deep learning // Journal of Big Data 6.1, 2019. – Р. 60.
Howard, Andrew G., et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications // arXiv preprint arXiv:1704.04861, 2017.
Wu, Yuhao, et al. Progressive Growing of GANs for Improved Quality, Stability, and Variation // arXiv preprint arXiv:1710.10196, 2017.
Howard, Andrew G., et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications // arXiv preprint arXiv:1704.04861, 2017.
Google AI Blog. Google supercharges machine learning tasks with TPU custom chip [Electronic resource] - URL: https://ai.googleblog.com/2016/05/announcing-tpu-tensor-processing-unit.html)
EIZO Rugged Solutions and NVIDIA Turing Push the Boundaries of Rugged AI [Electronic resource] - URL: https://www.unmannedsystemstechnology.com/wp-content/uploads/2021/01/NVIDIA_TuringGPU_whitepaper2021.pdf
