Regularizing Generative Adversarial Networks under Limited Data

Hung-Yu Tseng2Lu Jiang1Ce Liu1Ming-Hsuan Yang1,2,4Weilong Yang3

1Google Research   2UC Merced   3Waymo   4Yonsei University


Recent years have witnessed the rapid progress of generative adversarial networks (GANs). However, the success of the GAN models hinges on a large amount of training data. This work proposes a regularization approach for training robust GAN models on limited data. We theoretically show a connection between the regularized loss and an f-divergence called LeCam-divergence, which we find is more robust under limited training data. Extensive experiments on several benchmark datasets demonstrate that the proposed regularization scheme 1) improves the generalization performance and stabilizes the learning dynamics of GAN models under limited training data, and 2) complements the recent data augmentation methods. These properties facilitate training GAN models to achieve state-of-the-art performance when only limited training data of the ImageNet benchmark is available.


          author = {Tseng, Hung-Yu and Jiang, Lu and Liu, Ce and Yang, Ming-Hsuan and Yang, Weilong},
          title = {Regularing Generative Adversarial Networks under Limited Data},
          booktitle = {CVPR},
          year = {2021}
Page source: Po-Han Huang