Research map / Architectures and efficiency / RNNs and normalization
Revisiting Discriminator in GAN Compression: A Generator-discriminator Cooperative Compression Scheme
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 |
|---|---|---|---|
| add selective activation discriminator + GCC vs pruning with original discriminator | pruned generator · Cityscapes Pix2Pix | mIoU: 38.12 -> 42.88 | Helped |
| add selective activation discriminator + GCC vs pruning with original discriminator | pruned generator · Horse2zebra CycleGAN | FID: 145.1 -> 59.31 | Helped |
| add selective activation discriminator + GCC vs pruning with original discriminator | pruned generator · CelebA SAGAN | FID: 36.60 -> 25.21 | Helped |
| compress generator with GCC vs original uncompressed | SRGAN · SRGAN Set5 | PSNR: 29.88 -> 30.35 | Helped |
| add online collaborative distillation (ablation) | GCC · Cityscapes | mIoU: 39.17 -> 42.88 | Helped |
| add Texture loss to distillation (ablation) | GCC · Cityscapes | mIoU: 41.80 -> 42.88 | Helped |
| add MSE loss to distillation (ablation) | GCC · Cityscapes | mIoU: 40.07 -> 42.88 | Helped |
| use online instead of offline two-stage distillation (ablation) | GCC · Cityscapes | mIoU: 41.16 -> 42.88 | Helped |
| add discriminator-side distillation to generator-only distillation (ablation) | GCC · Cityscapes | mIoU: 42.31 -> 42.88 | Helped |
| add generator-side distillation to discriminator-only distillation (ablation) | GCC · Cityscapes | mIoU: 41.43 -> 42.88 | Helped |
| add student discriminator (vs discriminator-free distillation) (ablation) | GCC · Cityscapes | mIoU: 35.50 -> 42.88 | Helped |
| add selective activation to the discriminator (ablation) | GCC · Cityscapes | mIoU: 37.24 -> 42.88 | Helped |
Closest papers by meaning
- Discriminator-Cooperated Feature Map Distillation for GAN Compression (CVPR 2023 · similarity 0.94)
- GAN Compression: Efficient Architectures for Interactive Conditional GANs (CVPR 2020 · similarity 0.91)
- AutoGAN-Distiller: Searching to Compress Generative Adversarial Networks (ICML 2020 · similarity 0.90)
- Content-Aware GAN Compression (CVPR 2021 · similarity 0.90)
- Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold (AAAI 2024 · 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.