Research map / Image generation / GANs
AutoGAN-Distiller: Searching to Compress Generative Adversarial Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| AGD compression of CycleGAN (differentiable NAS with KD, FLOPs constraint) instead of CEC | CycleGAN · horse2zebra / zebra2horse / summer2winter / winter2summer | FID: 83.6 vs 96.15; 137.2 vs 157.9; 78.33 vs 78.58; 77.73 vs 79.26 | Helped |
| AGD compression of CycleGAN vs classical l1-norm structural pruning | CycleGAN · four unpaired translation tasks | FID: AGD far better than pruning (e.g., 83.6 vs 220.91 on horse2zebra) | Helped |
| 8-bit quantization applied on top of AGD-compressed CycleGAN (ablation) | AGD-compressed CycleGAN · horse2zebra, zebra2horse | FID: 85.74 vs 83.6 (horse2zebra); 140.08 vs 137.2 (zebra2horse) | Mixed |
| AGD compression of visualization-oriented ESRGAN vs structural pruning | ESRGAN · Set5/Set14/BSD100/Urban100 | PSNR: AGD 30.44/27.28/26.23/24.74 vs ESRGAN-Prune 28.07/25.21/24.74/22.67 | Helped |
| AGD compression of PSNR-oriented ESRGAN vs VDSR | ESRGAN (PSNR-oriented) · Set5/Set14/BSD100/Urban100 | PSNR: up to 0.44 dB better than VDSR with 84.1% fewer FLOPs and 32.6% smaller model | Helped |
| AGD compression of visualization-oriented ESRGAN | ESRGAN · Set5 / Set14 / BSD100 / Urban100 | PSNR: 30.44 vs 30.47; 27.28 vs 26.29 (up); 26.23 vs 25.32 (up); 24.74 vs 24.36 (up) | No clear effect |
Closest papers by meaning
- GAN Compression: Efficient Architectures for Interactive Conditional GANs (CVPR 2020 · similarity 0.91)
- Content-Aware GAN Compression (CVPR 2021 · similarity 0.91)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.88)
- Singular Value Scaling: Efficient Generative Model Compression via Pruned Weights Refinement (AAAI 2025 · similarity 0.88)
- Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold (AAAI 2024 · similarity 0.87)
- The GAN is dead; long live the GAN! A Modern GAN Baseline (NeurIPS 2024 · 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.