AI Paper Atlas

Research map / Image generation / GANs

Masked Generative Adversarial Networks are Data-Efficient Generation Learners

Jiaxing Huang, Kaiwen Cui, Dayan Guan, Aoran Xiao, Fangneng Zhan, Shijian Lu, Shengcai Liao, Eric Xing

NeurIPS 2022 · Image generation › GANs · approach: Stabilized GAN training · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add spatial masking (ablation)BigGAN · CIFAR-10 10% dataFID: 48.08 -> 28.35Improved
add random shift on top of spatial masking (ablation)BigGAN · CIFAR-10 10% dataFID: 28.35 -> 22.56Improved
add basic spectral masking with equal probabilities (ablation)BigGAN · CIFAR-10 10% dataFID: 48.08 -> 35.20Improved
use self-adaptive probabilities for balanced spectral masking (ablation)BigGAN · CIFAR-10 10% dataFID: 35.20 -> 28.78Improved
combine shifted spatial and balanced spectral masking (ablation)BigGAN · CIFAR-10 10% dataFID: 22.56 -> 15.89Improved
add MaskedGANBigGAN · CIFAR-10 100% dataFID: 9.74 ± 0.06 -> 8.41 ± 0.03Improved
add MaskedGANBigGAN · CIFAR-100 100% dataFID: 13.60 ± 0.07 -> 11.65 ± 0.03No clear effect
add MaskedGANBigGAN · CIFAR-100 10% dataFID: 66.71 ± 0.01 -> 24.02 ± 0.12Improved
add MaskedGANBigGAN · ImageNet 2.5% dataFID: 133.80 ± 0.76 -> 38.62 ± 0.37Improved
add MaskedGANStyleGAN-v2 · 100-shot PandaFID: 34.27 -> 8.93 ± 0.06Improved
add MaskedGANStyleGAN-v2 · FFHQ 1.25% dataFID: 127.82 ± 0.58 -> 44.36 ± 0.24Improved
add MaskedGANStyleGAN-v2 · AFHQ-Cat 20% dataFID: 12.51 ± 0.09 -> 12.03 ± 0.10No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy