Investigating the effect of loss functions on single-image GAN performance

Eyyüp Yıldız; Erkan Yüksel; Selçuk Sevgen

doi:10.38088/jise.1497968

EN

Investigating the effect of loss functions on single-image GAN performance

Abstract

Loss functions are crucial in training generative adversarial networks (GANs) and shaping the resulting outputs. These functions, specifically designed for GANs, optimize generator and discriminator networks together but in opposite directions. GAN models, which typically handle large datasets, have been successful in the field of deep learning. However, exploring the factors that influence the success of GAN models developed for limited data problems is an important area of research. In this study, we conducted a comprehensive investigation into the loss functions commonly used in GAN literature, such as binary cross entropy (BCE), Wasserstein generative adversarial network (WGAN), least squares generative adversarial network (LSGAN), and hinge loss. Our research focused on examining the impact of these loss functions on improving output quality and ensuring training convergence in single-image GANs. Specifically, we evaluated the performance of a single-image GAN model, SinGAN, using these loss functions in terms of image quality and diversity. Our experimental results demonstrated that loss functions successfully produce high-quality, diverse images from a single training image. Additionally, we found that the WGAN-GP and LSGAN-GP loss functions are more effective for single-image GAN models.

Keywords

Ethical Statement

Etik iznine gerek yoktur.

References

[1] Shahriar, S. (2022). GAN computers generate arts? A survey on visual arts, music, and literary text generation using generative adversarial network. Displays, 73, 102237.
[2] Chakraborty, T., KS, U. R., Naik, S. M., Panja, M., & Manvitha, B. (2024). Ten years of generative adversarial nets (GANs): a survey of the state-of-the-art. Machine Learning: Science and Technology, 5(1), 011001.
[3] Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville A. & Bengio, Y. (2020). Generative adversarial networks. Communications of the ACM, 63(11), 139-144.
[4] Salimans, T., Goodfellow, I., Zaremba, W., Cheung, V., Radford, A., & Chen, X. (2016). Improved techniques for training gans. Advances in neural information processing systems, 29.
[5] Zhang, Z., Li, M., & Yu, J. (2018). On the convergence and mode collapse of GAN. SIGGRAPH Asia 2018 Technical Briefs, 1-4.
[6] Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B. & Hochreiter, S. (2017). Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30.
[7] Iglesias, G., Talavera, E. & Díaz-Álvarez, A. (2023). A survey on GANs for computer vision: Recent research, analysis and taxonomy. Computer Science Review, 48, 100553.
[8] Xia, W., Zhang, Y., Yang, Y., Xue, J. H., Zhou, B., & Yang, M. H. (2022). Gan inversion: A survey. IEEE transactions on pattern analysis and machine intelligence, 45(3), 3121-3138.

[9] Wang, P., Li, Y., Singh, K. K., Lu, J. & Vasconcelos, N. (2021). Imagine: Image synthesis by image-guided model inversion. IEEE/CVF Conference on Computer Vision and Pattern Recognition, 3681-3690.
[10] Yildiz, E., Yuksel, M. E., & Sevgen, S. (2024). A Single-Image GAN Model Using Self-Attention Mechanism and DenseNets. Neurocomputing, 596, 127873.
[11] Zhang, Z., Han, C. & Guo, T. (2021). Exsingan: Learning an explainable generative model from a single image. 32nd British Machine Vision Conference.
[12] Shaham, T. R., Dekel, T. & Michaeli, T. (2019). Singan: Learning a generative model from a single natural image. IEEE/CVF international conference on computer vision, 4570-4580.
[13] Ulyanov, D., Vedaldi, A. ve Lempitsky, V. (2018). Deep image prior. IEEE conference on computer vision and pattern recognition, 9446-9454.
[14] Shocher, A., Bagon, S., Isola, P., & Irani, M. (2019). Ingan: Capturing and retargeting the" dna" of a natural image. IEEE/CVF international conference on computer vision, 4492-4501.
[15] Isola, P., Zhu, J. Y., Zhou, T. & Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. IEEE conference on computer vision and pattern recognition, 1125-1134.
[16] Hinz, T., Fisher, M., Wang, O. & Wermter, S. (2021). Improved techniques for training single-image gans. IEEE/CVF Winter Conference on Applications of Computer Vision, 1300-1309.
[17] Karras, T., Aila, T., Laine, S. & Lehtinen, J. (2018). Progressive Growing of GANs for Improved Quality, Stability, and Variation. International Conference on Learning Representations.
[18] Granot, N., Feinstein, B., Shocher, A., Bagon, S. ve Irani, M. (2022). Drop the gan: In defense of patches nearest neighbors as single image generative models. IEEE/CVF Conference on Computer Vision and Pattern Recognition, 13460-13469.
[19] Arjovsky M., Chintala S. & Bottou L. (2017). Wasserstein generative adversarial networks. 34th International Conference on Machine Learning, ICML, 298–321.
[20] Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V. & Courville, A. C. (2017). Improved training of wasserstein gans. Advances in neural information processing systems, 30.
[21] Mao, X., Li, Q., Xie, H., Lau, R. Y., Wang, Z., & Paul Smolley, S. (2017). Least squares generative adversarial networks. IEEE international conference on computer vision, 2794-2802.
[22] Lim, J. H., & Ye, J. C. (2017). Geometric gan. arXiv preprint arXiv:1705.02894.
[23] Iglesias, G., Talavera, E. & Díaz-Álvarez, A. (2023). A survey on GANs for computer vision: Recent research, analysis and taxonomy. Computer Science Review, 48, 100553.
[24] Jabbar, A., Li, X., & Omar, B. (2021). A survey on generative adversarial networks: Variants, applications, and training. ACM Computing Surveys (CSUR), 54(8), 1-49.
[25] Wang, Z., Simoncelli, E. P. & Bovik, A. C. (2003). Multiscale structural similarity for image quality assessment. The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers IEEE, 1398-1402.
[26] Zhang, R., Isola, P., Efros, A. A., Shechtman, E. & Wang, O. (2018). The unreasonable effectiveness of deep features as a perceptual metric. IEEE conference on computer vision and pattern recognition, 586-595.

Details

Primary Language

English

Subjects

Computer Vision , Pattern Recognition , Deep Learning , Neural Networks , Machine Learning (Other)

Journal Section

Research Article

Authors

Eyyüp Yıldız ^*
0000-0002-7051-3368
Türkiye

Erkan Yüksel
0000-0001-8976-9964
Türkiye

Selçuk Sevgen
0000-0003-1443-1779
Türkiye

Early Pub Date

December 11, 2024

Publication Date

December 31, 2024

Submission Date

June 8, 2024

Acceptance Date

August 9, 2024

Published in Issue

Year 2024 Volume: 8 Number: 2

DOI

https://doi.org/10.38088/jise.1497968

IZ

https://izlik.org/JA85XD26UP

Cite

RIS / Bibtex

APA

Yıldız, E., Yüksel, E., & Sevgen, S. (2024). Investigating the effect of loss functions on single-image GAN performance. Journal of Innovative Science and Engineering, 8(2), 213-225. https://doi.org/10.38088/jise.1497968

AMA

1.Yıldız E, Yüksel E, Sevgen S. Investigating the effect of loss functions on single-image GAN performance. JISE. 2024;8(2):213-225. doi:10.38088/jise.1497968

Chicago

Yıldız, Eyyüp, Erkan Yüksel, and Selçuk Sevgen. 2024. “Investigating the Effect of Loss Functions on Single-Image GAN Performance”. Journal of Innovative Science and Engineering 8 (2): 213-25. https://doi.org/10.38088/jise.1497968.

EndNote

Yıldız E, Yüksel E, Sevgen S (December 1, 2024) Investigating the effect of loss functions on single-image GAN performance. Journal of Innovative Science and Engineering 8 2 213–225.

IEEE

[1]E. Yıldız, E. Yüksel, and S. Sevgen, “Investigating the effect of loss functions on single-image GAN performance”, JISE, vol. 8, no. 2, pp. 213–225, Dec. 2024, doi: 10.38088/jise.1497968.

ISNAD

Yıldız, Eyyüp - Yüksel, Erkan - Sevgen, Selçuk. “Investigating the Effect of Loss Functions on Single-Image GAN Performance”. Journal of Innovative Science and Engineering 8/2 (December 1, 2024): 213-225. https://doi.org/10.38088/jise.1497968.

JAMA

1.Yıldız E, Yüksel E, Sevgen S. Investigating the effect of loss functions on single-image GAN performance. JISE. 2024;8:213–225.

MLA

Yıldız, Eyyüp, et al. “Investigating the Effect of Loss Functions on Single-Image GAN Performance”. Journal of Innovative Science and Engineering, vol. 8, no. 2, Dec. 2024, pp. 213-25, doi:10.38088/jise.1497968.

Vancouver

1.Eyyüp Yıldız, Erkan Yüksel, Selçuk Sevgen. Investigating the effect of loss functions on single-image GAN performance. JISE. 2024 Dec. 1;8(2):213-25. doi:10.38088/jise.1497968

Cited By

Advanced deep learning for PET image reconstruction and quality improvement

The Imaging Science Journal

https://doi.org/10.1080/13682199.2025.2558374

Investigating the effect of loss functions on single-image GAN performance

Abstract

Keywords

Öz

Ethical Statement

References

Details

Primary Language

Subjects

Journal Section

Authors

Early Pub Date

Publication Date

Submission Date

Acceptance Date

Published in Issue

DOI

IZ

Cite

Cited By

Advanced deep learning for PET image reconstruction and quality improvement