Spectral Normalization for Generative Adversarial Networks

Code

• github.com/pfnet-research/sngan_projection
  OfficialIn papertf★ 0
• github.com/open-mmlab/mmgeneration
  pytorch★ 2,016
• github.com/alhasan-abdellatif/cgans
  pytorch★ 7
• github.com/karolrogozinski/cern_alice_fast_sim_corrvae
  pytorch★ 3
• github.com/crcrpar/pytorch.sngan_projection
  In paperpytorch★ 0
• github.com/hinofafa/Self-Attention-GAN
  pytorch★ 0
• github.com/ksagit/sn_gan_pytorch
  pytorch★ 0
• github.com/Xiaoming-Yu/SingleGAN
  pytorch★ 0
• github.com/yanqi1811/self-attention
  pytorch★ 0
• github.com/hanyoseob/pytorch-sngan
  pytorch★ 0

Abstract

One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs (SN-GANs) is capable of generating images of better or equal quality relative to the previous training stabilization techniques.

Tasks

Benchmark Results

Reproductions