Papers › Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks...

Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks via Singular Values

3 Mar 2022CVPR 2022 1arXiv:2203.01993archive 2025-07-28

Ahmed Imtiaz Humayun, Randall Balestriero, Richard Baraniuk

We present Polarity Sampling, a theoretically justified plug-and-play method for controlling the generation quality and diversity of pre-trained deep generative networks DGNs). Leveraging the fact that DGNs are, or can be approximated by, continuous piecewise affine splines, we derive the analytical DGN output space distribution as a function of the product of the DGN's Jacobian singular values raised to a power ρ. We dub ρ the polarity parameter and prove that ρ focuses the DGN sampling on the modes (ρ< 0) or anti-modes (ρ> 0) of the DGN output-space distribution. We demonstrate that nonzero polarity values achieve a better precision-recall (quality-diversity) Pareto frontier than standard methods, such as truncation, for a number of state-of-the-art DGNs. We also present quantitative and qualitative results on the improvement of overall generation quality (e.g., in terms of the Frechet Inception Distance) for a number of state-of-the-art DGNs, including StyleGAN3, BigGAN-deep, NVAE, for different conditional and unconditional image generation tasks. In particular, Polarity Sampling redefines the state-of-the-art for StyleGAN2 on the FFHQ Dataset to FID 2.57, StyleGAN2 on the LSUN Car Dataset to FID 2.27 and StyleGAN3 on the AFHQv2 Dataset to FID 3.95. Demo: bit.ly/polarity-samp

PaperPDFConference PDFCode

In Syntology View this paper on Syntology: its repositories, every harvested function with whether it ran, its licence and the call to fetch it.

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

Code

AhmedImtiazPrio/magnet-polarity officialmentioned in papermentioned on GitHubtf 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

DiversityImage GenerationUnconditional Image Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Generation AFHQV2 Polarity-StyleGAN3 FID 3.95 #1 of 7 Archive leaderboard report
Image Generation CelebA-HQ 1024x1024 Polarity-ProGAN FID 7.28 #6 of 10 Archive leaderboard report
Image Generation FFHQ 1024 x 1024 Polarity-StyleGAN2 FID 2.57 #3 of 20 Archive leaderboard report
Image Generation ImageNet 256x256 Polarity-BigGAN FID 6.82 #91 of 94 Archive leaderboard report
Image Generation LSUN Car 512 x 384 Polarity-StyleGAN2 FID 2.27 #1 of 2 Archive leaderboard report
Image Generation LSUN Cat 256 x 256 Polarity-StyleGAN2 FID 6.34 #4 of 8 Archive leaderboard report
Image Generation LSUN Churches 256 x 256 Polarity-StyleGAN2 FID 3.92 #11 of 27 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 ConvolutionAdamAverage PoolingBatch NormalizationBigGAN-deepBottleneck Residual BlockConditional Batch NormalizationConvolutionDense ConnectionsDepthwise ConvolutionDepthwise Separable ConvolutionEarly StoppingFeedforward NetworkGAN Hinge LossLinear LayerNVAENVAE Encoder Residual CellNVAE Generative Residual CellNon-Local BlockNon-Local OperationNormalizing FlowsOff-Diagonal Orthogonal RegularizationPath Length RegularizationPointwise ConvolutionProjection DiscriminatorR1 RegularizationReLUResidual ConnectionResidual Normal DistributionSAGANSigmoid ActivationSoftmaxSpectral NormalizationSqueeze-and-Excitation BlockTTURTruncation TrickWeight DemodulationWeight Normalization

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