Papers › Adaptive Weighted Discriminator for Training Generative Adversarial Networks

Adaptive Weighted Discriminator for Training Generative Adversarial Networks

5 Dec 2020CVPR 2021 1arXiv:2012.03149archive 2025-07-28

Vasily Zadorozhnyy, Qiang Cheng, Qiang Ye

Generative adversarial network (GAN) has become one of the most important neural network models for classical unsupervised machine learning. A variety of discriminator loss functions have been developed to train GAN's discriminators and they all have a common structure: a sum of real and fake losses that only depends on the actual and generated data respectively. One challenge associated with an equally weighted sum of two losses is that the training may benefit one loss but harm the other, which we show causes instability and mode collapse. In this paper, we introduce a new family of discriminator loss functions that adopts a weighted sum of real and fake parts, which we call adaptive weighted loss functions or aw-loss functions. Using the gradients of the real and fake parts of the loss, we can adaptively choose weights to train a discriminator in the direction that benefits the GAN's stability. Our method can be potentially applied to any discriminator model with a loss that is a sum of the real and fake parts. Experiments validated the effectiveness of our loss functions on an unconditional image generation task, improving the baseline results by a significant margin on CIFAR-10, STL-10, and CIFAR-100 datasets in Inception Scores and FID.

PaperPDFConference PDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

vasily789/adaptive-weighted-gans officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Conditional Image GenerationImage GenerationUnconditional Image Generation

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Conditional Image Generation CIFAR-10 aw-BigGAN FID 6.89 #8 of 25 Archive leaderboard report
Conditional Image Generation CIFAR-10 aw-BigGAN Inception score 9.52 #8 of 25 Archive leaderboard report
Conditional Image Generation CIFAR-10 aw-SN-GAN FID 8.03 #10 of 25 Archive leaderboard report
Conditional Image Generation CIFAR-10 aw-SN-GAN Inception score 9 #10 of 25 Archive leaderboard report
Conditional Image Generation CIFAR-100 aw-BigGAN FID 10.23 #5 of 7 Archive leaderboard report
Conditional Image Generation CIFAR-100 aw-BigGAN Inception Score 11.22 #5 of 7 Archive leaderboard report
Conditional Image Generation CIFAR-100 aw-SN-GAN FID 14 #6 of 7 Archive leaderboard report
Conditional Image Generation CIFAR-100 aw-SN-GAN Inception Score 9.48 #6 of 7 Archive leaderboard report
Image Generation CIFAR-100 aw-AutoGAN FID 19 #6 of 9 Archive leaderboard report
Image Generation CIFAR-100 aw-AutoGAN Inception Score 8.9 #6 of 9 Archive leaderboard report
Image Generation CIFAR-100 aw-SN-GAN FID 19.08 #7 of 9 Archive leaderboard report
Image Generation CIFAR-100 aw-SN-GAN Inception Score 8.31 #7 of 9 Archive leaderboard report
Image Generation STL-10 aw-AutoGAN FID 26.32 #20 of 31 Archive leaderboard report
Image Generation STL-10 aw-AutoGAN Inception score 9.59 #20 of 31 Archive leaderboard report
Image Generation STL-10 aw-SN-GAN FID 34.72 #23 of 31 Archive leaderboard report
Image Generation STL-10 aw-SN-GAN Inception score 9.61 #23 of 31 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

1x1 ConvolutionAdamAutoGANBatch NormalizationBigGANConditional Batch NormalizationConvolutionDense ConnectionsEarly StoppingFeedforward NetworkGAN Hinge LossLinear LayerNon-Local BlockNon-Local OperationOff-Diagonal Orthogonal RegularizationProjection DiscriminatorReLUResidual BlockResidual ConnectionSAGANSNGANSoftmaxSpectral NormalizationTTURTruncation Trick

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections