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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add spatial masking (ablation) | BigGAN · CIFAR-10 10% data | FID: 48.08 -> 28.35 | Improved |
| add random shift on top of spatial masking (ablation) | BigGAN · CIFAR-10 10% data | FID: 28.35 -> 22.56 | Improved |
| add basic spectral masking with equal probabilities (ablation) | BigGAN · CIFAR-10 10% data | FID: 48.08 -> 35.20 | Improved |
| use self-adaptive probabilities for balanced spectral masking (ablation) | BigGAN · CIFAR-10 10% data | FID: 35.20 -> 28.78 | Improved |
| combine shifted spatial and balanced spectral masking (ablation) | BigGAN · CIFAR-10 10% data | FID: 22.56 -> 15.89 | Improved |
| add MaskedGAN | BigGAN · CIFAR-10 100% data | FID: 9.74 ± 0.06 -> 8.41 ± 0.03 | Improved |
| add MaskedGAN | BigGAN · CIFAR-100 100% data | FID: 13.60 ± 0.07 -> 11.65 ± 0.03 | No clear effect |
| add MaskedGAN | BigGAN · CIFAR-100 10% data | FID: 66.71 ± 0.01 -> 24.02 ± 0.12 | Improved |
| add MaskedGAN | BigGAN · ImageNet 2.5% data | FID: 133.80 ± 0.76 -> 38.62 ± 0.37 | Improved |
| add MaskedGAN | StyleGAN-v2 · 100-shot Panda | FID: 34.27 -> 8.93 ± 0.06 | Improved |
| add MaskedGAN | StyleGAN-v2 · FFHQ 1.25% data | FID: 127.82 ± 0.58 -> 44.36 ± 0.24 | Improved |
| add MaskedGAN | StyleGAN-v2 · AFHQ-Cat 20% data | FID: 12.51 ± 0.09 -> 12.03 ± 0.10 | No clear effect |
Closest papers by meaning
- Training Generative Adversarial Networks with Limited Data (NeurIPS 2020 · similarity 0.90)
- Masked Auto-Encoders Meet Generative Adversarial Networks and Beyond (CVPR 2023 · similarity 0.90)
- Large Scale GAN Training for High Fidelity Natural Image Synthesis (ICLR 2019 · similarity 0.90)
- Projected GANs Converge Faster (NeurIPS 2021 · similarity 0.88)
- Spider GAN: Leveraging Friendly Neighbors To Accelerate GAN Training (CVPR 2023 · similarity 0.88)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.87)
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.
