Timezone: »

 
Poster
Deceive D: Adaptive Pseudo Augmentation for GAN Training with Limited Data
Liming Jiang · Bo Dai · Wayne Wu · Chen Change Loy

Tue Dec 07 04:30 PM -- 06:00 PM (PST) @ Virtual
in Poster Session 2 »

Generative adversarial networks (GANs) typically require ample data for training in order to synthesize high-fidelity images. Recent studies have shown that training GANs with limited data remains formidable due to discriminator overfitting, the underlying cause that impedes the generator's convergence. This paper introduces a novel strategy called Adaptive Pseudo Augmentation (APA) to encourage healthy competition between the generator and the discriminator. As an alternative method to existing approaches that rely on standard data augmentations or model regularization, APA alleviates overfitting by employing the generator itself to augment the real data distribution with generated images, which deceives the discriminator adaptively. Extensive experiments demonstrate the effectiveness of APA in improving synthesis quality in the low-data regime. We provide a theoretical analysis to examine the convergence and rationality of our new training strategy. APA is simple and effective. It can be added seamlessly to powerful contemporary GANs, such as StyleGAN2, with negligible computational cost. Code: https://github.com/EndlessSora/DeceiveD.

Keywords: Generative Model

Author Information

Liming Jiang (Nanyang Technological University)
Bo Dai (Shanghai AI Lab)
Wayne Wu (Tsinghua University)
Chen Change Loy (Nanyang Technological University)

More from the Same Authors