Gradient Normalization for Generative Adversarial Networks
Authors:
- 1 October 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 6353-6362
Abstract
In this paper, we propose a novel normalization method called gradient normalization (GN) to tackle the training instability of Generative Adversarial Networks (GANs) caused by the sharp gradient space. Unlike existing work such as gradient penalty and spectral normalization, the proposed GN only imposes a hard 1-Lipschitz constraint on the discriminator function, which increases the capacity of the discriminator. Moreover, the proposed gradient normalization can be applied to different GAN architectures with little modification. Extensive experiments on four datasets show that GANs trained with gradient normalization outperform existing methods in terms of both Frechet Inception Distance and Inception Score.Keywords
All Related Versions
Funding Information
- Ministry of Science and Technology
This publication has 8 references indexed in Scilit:
- Off-Policy Reinforcement Learning for Efficient and Effective GAN Architecture Search2020
- Spectral Regularization for Combating Mode Collapse in GANs2019
- AutoGAN: Neural Architecture Search for Generative Adversarial Networks2019
- MirrorGAN: Learning Text-To-Image Generation by Redescription2019
- Wasserstein Divergence for GANs2018
- Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks2017
- Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network2017
- 80 Million Tiny Images: A Large Data Set for Nonparametric Object and Scene RecognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008