Training GANs with Stronger Augmentations via Contrastive Discriminator

By Jongheon Jeong, Jinwoo Shin. 2021

70 citations recorded by the source corpus.

Recent works in Generative Adversarial Networks (GANs) are actively revisiting various data augmentation techniques as an effective way to prevent discriminator overfitting. It is still unclear, however, that which augmentations could actually improve GANs, and in particular, how to apply a wider range of augmentations in training. In this paper, we propose a novel way to address these questions by incorporating a recent contrastive representation learning scheme into the GAN discriminator, coined ContraD. This"fusion"enables the discriminators to work with much stronger augmentations without increasing their training instability, thereby preventing the discriminator overfitting issue in GANs more effectively. Even better, we observe that the contrastive learning itself also benefits from our GAN training, i.e., by maintaining discriminative features between real and fake samples, suggesting a strong coherence between the two worlds: good contrastive representations are also good for GAN discriminators, and vice versa. Our experimental results show that GANs with ContraD consistently improve FID and IS compared to other recent techniques incorporating data augmentations, still maintaining highly discriminative features in the discriminator in terms of the linear evaluation. Finally, as a byproduct, we also show that our GANs trained in an unsupervised manner (without labels) can induce many conditional generative models via a simple latent sampling, leveraging the learned features of ContraD. Code is available at https://github.com/jh-jeong/ContraD.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Generative Models, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Bolei Zhou
2. Mario Fritz
3. Larry S. Davis
4. Yujun Shen
5. Ceyuan Yang
6. Yinghao Xu
7. Ning Yu
8. Aysegul Dundar

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Dacheng Tao
3. Ming-Hsuan Yang
4. Trevor Darrell
5. Bolei Zhou
6. Chen Change Loy
7. Xiaogang Wang
8. Bernt Schiele

In Generative Models

1. Yujun Shen
2. Bolei Zhou
3. Yinghao Xu
4. Ceyuan Yang
5. Mario Fritz
6. Ning Yu
7. Eli Shechtman
8. Andrew Tao

In Signal Processing

1. Bolei Zhou
2. Yinghao Xu
3. Larry S. Davis
4. Mario Fritz
5. Ning Yu
6. Yujun Shen
7. Bryan Catanzaro
8. Andrew Tao

Citations in the corpus

• Towards Language-Free Training for Text-to-Image Generation (2021, neutral): “Similar ideas have been exploited in [15, 17, 56].”
• Dual Contrastive Loss and Attention for GANs (2021, background): “It is orthogonal to [40, 106, 36, 47] where their contrastive losses serve only as an incremental auxiliary to the conventional adversarial loss, apply to the generator rather than the discriminator, and/or require expensive class annotations or augmentation for generation.”
• Data-Efficient Instance Generation from Instance Discrimination (2021, background): “Akin to supervised contrastive loss [27], some concurrent work [20, 21, 43] reformulates the conventional bi-classification task (i.”