for Data-Efficient GAN Training Shengyu Zhao 1,2 Ji Lin 1 Jun-Yan Zhu - - PowerPoint PPT Presentation
Differentiable Augmentation for Data-Efficient GAN Training Shengyu Zhao 1,2 Ji Lin 1 Jun-Yan Zhu 3,4 Song Han 1 Zhijian Liu 1 1 MIT 2 IIIS, Tsinghua University 3 Adobe Research 4 CMU NeurIPS 2020 Data Is Expensive Computation Algorithm
1MIT 2IIIS, Tsinghua University 3Adobe Research 4CMU
NeurIPS 2020
Shengyu Zhao1,2 Zhijian Liu1 Ji Lin1 Song Han1 Jun-Yan Zhu3,4
Computation Algorithm Computation Algorithm
FFHQ dataset: 70,000 selective post-processed human faces
ImageNet dataset: millions of images from diverse categories
Big Data
Obama 100 images
Generated samples of StyleGAN2 (Karras et al.) using only hundreds of images
Cat (Simard et al.) 160 images Dog (Simard et al.) 389 images
11.1 23.1 36.0 5 10 15 20 25 30 35 40
100% training data 20% training data 10% training data
CIFAR-10
Artifacts from Color jittering Artifacts from Translation Artifacts from Cutout (DeVries et al.)
Generated images
Color Translation Cutout Color Translation Cutout
fakes reals
11.1 23.1 36.0 9.9 12.2 14.5 5 10 15 20 25 30 35 40
100% training data 20% training data 10% training data
11.1 23.1 36.0 9.9 12.2 14.5 5 10 15 20 25 30 35 40
100% training data 20% training data 10% training data
CIFAR-10
Obama 100 images Cat (Simard et al.) 160 images Dog (Simard et al.) 389 images
10 20 30 40 50 60 Performance FID↓
Scale/Shift (Noguchi et al.) MineGAN (Wang et al.) TransferGAN (Wang et al.) FreezeD (Mo et al.) Ours
1 10 100 1000 10000 100000 Data # Training Images
No pre-training
100-shot Obama