Research map / Image generation / GANs
F-Drop&Match: GANs With a Dead Zone in the High-Frequency Domain
Shin'ya Yamaguchi, Sekitoshi Kanai
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 F-Drop and F-Match together | SNGAN · CIFAR-10 | FID: 14.3 -> 10.7 | Improved |
| add F-Drop and F-Match together | SNGAN · CIFAR-100 | FID: 15.2 -> 13.8 | Improved |
| add F-Drop and F-Match together | SNGAN · TinyImageNet | FID: 21.8 -> 18.9 | Improved |
| add F-Drop and F-Match together | SNGAN · CelebA | FID: 7.98 -> 6.78 | Improved |
| add F-Drop and F-Match together | SNGAN · ImageNet | FID: 62.5 -> 60.4 | Improved |
| add F-Match to F-Drop (ablation) | F-Drop · CIFAR-10 | FID: 14.1 -> 10.7 | Improved |
| add F-Drop alone (ablation) | SNGAN · CIFAR-10 | FID: 14.3 -> 14.1 | No clear effect |
| add F-Drop alone (ablation) | SNGAN · CIFAR-10 | frequency gap: 6.89 -> 5.94 | Improved |
| add F-Drop alone (ablation) | SNGAN · CelebA | frequency gap: 4.49 -> 4.60 | Hurt |
| add F-Drop&Match | SNGAN · CIFAR-100 | frequency gap: 7.01 -> 4.16 | Improved |
| replace Binomial upsampling baseline: add Binomial | SNGAN · CIFAR-10 | FID: 14.3 -> 35.9 | Hurt |
| add F-Drop&Match to StyleGAN2-ADA | StyleGAN2-ADA · AFHQ-Wild | FID: 3.05 -> 2.62 | Improved |
Closest papers by meaning
- FreGAN: Exploiting Frequency Components for Training GANs under Limited Data (NeurIPS 2022 · similarity 0.89)
- On the Frequency Bias of Generative Models (NeurIPS 2021 · similarity 0.89)
- Spatial Frequency Bias in Convolutional Generative Adversarial Networks (AAAI 2022 · similarity 0.87)
- Frequency-Aware Deepfake Detection: Improving Generalizability through Frequency Space Domain Learning (AAAI 2024 · similarity 0.87)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.86)
- SMaRt: Improving GANs with Score Matching Regularity (ICML 2024 · similarity 0.86)
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.
