Freeze the Discriminator: a Simple Baseline for Fine-Tuning GANs

Abstract

Generative adversarial networks (GANs) have shown outstanding performance on a wide range of problems in computer vision, graphics, and machine learning, but often require numerous training data and heavy computational resources. To tackle this issue, several methods introduce a transfer learning technique in GAN training. They, however, are either prone to overfitting or limited to learning small distribution shifts. In this paper, we show that simple fine-tuning of GANs with frozen lower layers of the discriminator performs surprisingly well. This simple baseline, FreezeD, significantly outperforms previous techniques used in both unconditional and conditional GANs. We demonstrate the consistent effect using StyleGAN and SNGAN-projection architectures on several datasets of Animal Face, Anime Face, Oxford Flower, CUB-200-2011, and Caltech-256 datasets. The code and results are available at https://github.com/sangwoomo/FreezeD.

Tech report; High resolution images are in https://github.com/sangwoomo/FreezeD

References in corpus (6)

• Distilling the Knowledge in a Neural Network
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• How transferable are features in deep neural networks?
• Small-GAN: Speeding Up GAN Training Using Core-sets
• Learning What and Where to Transfer
• Mining GOLD Samples for Conditional GANs

Cited by in corpus (8)

• Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis
• StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation
• Reinforcing Generated Images via Meta-learning for One-Shot Fine-Grained Visual Recognition
• Few-shot Image Generation via Cross-domain Correspondence
• Unsupervised Image-to-Image Translation via Pre-trained StyleGAN2 Network
• Regularizing Generative Adversarial Networks under Limited Data
• StyleUV: Diverse and High-fidelity UV Map Generative Model
• Fine-Tuning StyleGAN2 For Cartoon Face Generation