EUVIMEDEUVIMED entdecken
Crossref

A new formulation of Lipschitz constrained with functional gradient learning for GANs

Chang Wan, Ke Fan, Xinwei Sun, Yanwei Fu, Minglu Li, Yunliang Jiang, Zhonglong Zheng

Machine Learning · 2025-03 · 114 (3)

DOI: 10.1007/s10994-024-06633-9 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Machine Learning, online erschienen am 04.02.2025. Verfasst von Chang Wan und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 42 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Zhejiang Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper introduces a promising alternative method for training Generative Adversarial Networks (GANs) on large-scale datasets with clear theoretical guarantees. GANs are typically learned through a minimax game between a generator and a discriminator, which is known to be empirically unstable. Previous learning paradigms have encountered mode collapse issues without a theoretical solution. To address these challenges, we propose a novel Lipschitz-constrained Functional Gradient GANs learning (Li-CFG) method to stabilize the training of GAN and provide a theoretical foundation for effectively increasing the diversity of synthetic samples by reducing the neighborhood size of the latent vector. Specifically, we demonstrate that the neighborhood size of the latent vector can be reduced by increasing the norm of the discriminator gradient, resulting in enhanced diversity of synthetic samples. To efficiently enlarge the norm of the discriminator gradient, we introduce a novel ε\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varvec{\varepsilon }}$$\end{document}-centered gradient penalty that amplifies the norm of the discriminator gradient using the hyper-parameter ε\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varvec{\varepsilon }}$$\end{document}. In comparison to other constraints, our method enlarging the discriminator norm, thus obtaining the smallest neighborhood size of the latent vector. Extensive experiments on benchmark datasets for image generation demonstrate the efficacy of the Li-CFG method and the ε\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varvec{\varepsilon }}$$\end{document}-centered gradient penalty. The results showcase improved stability and increased diversity of synthetic samples.

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Zhejiang Province

Zitieren

APA

Wan, C., Fan, K., Sun, X., Fu, Y., Li, M., Jiang, Y., & Zheng, Z. (2025). A new formulation of Lipschitz constrained with functional gradient learning for GANs. Machine Learning, 114(3). https://doi.org/10.1007/s10994-024-06633-9

Vancouver

Wan C, Fan K, Sun X, Fu Y, Li M, Jiang Y, et al. A new formulation of Lipschitz constrained with functional gradient learning for GANs. Machine Learning. 2025;114(3). doi:10.1007/s10994-024-06633-9

BibTeX
@article{Wan2025,
  author = {Wan, Chang and Fan, Ke and Sun, Xinwei and Fu, Yanwei and Li, Minglu and Jiang, Yunliang and Zheng, Zhonglong},
  title = {A new formulation of Lipschitz constrained with functional gradient learning for GANs},
  journal = {Machine Learning},
  year = {2025},
  volume = {114},
  number = {3},
  doi = {10.1007/s10994-024-06633-9}
}
Kostenlos starten →