Research map / Image generation / GANs
Generator Knows What Discriminator Should Learn in Unconditional GANs
Gayoung Lee, Hyunsu Kim, Junho Kim, Seonghyeon Kim, Jung-Woo Ha, Yunjey Choi
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 GGDR (generator-feature-map supervision of discriminator) | StyleGAN2 · FFHQ full | FID: 3.66 -> 3.14 | Improved |
| add GGDR | StyleGAN2 · FFHQ 2k | FID: 78.80 -> 70.59 | Improved |
| add GGDR | ADA · FFHQ 2k | FID: 16.49 -> 18.28 | Hurt |
| add GGDR | ADA · FFHQ 10k | FID: 8.29 -> 6.11 | Improved |
| add GGDR | ADA · CIFAR-10 | FID: 2.92 -> 2.15 | Improved |
| add GGDR | ADA · CIFAR-10 | IS: 9.83 -> 10.02 | Improved |
| add GGDR | StyleGAN2 · LSUN Cat | FID: 7.98 -> 5.28 | Improved |
| add GGDR | StyleGAN2 · LSUN Cat | Recall: 0.27 -> 0.38 | Improved |
| add GGDR | StyleGAN2 · FFHQ | Precision: 0.69 -> 0.69 | No clear effect |
| add GGDR | ADA · AFHQ Dog | FID: 7.40 -> 4.59 | Improved |
| add GGDR | ADA · Landscape | Precision: 0.72 -> 0.69 | Hurt |
| increase guidance feature map size from none to 64x64 (ablation) | StyleGAN2+GGDR · LSUN Cat | FID: 7.98 -> 5.28 | Improved |
Closest papers by meaning
- Return of Unconditional Generation: A Self-supervised Representation Generation Method (NeurIPS 2024 · similarity 0.90)
- OSSGAN: Open-Set Semi-Supervised Image Generation (CVPR 2022 · similarity 0.89)
- Self-Supervised GANs via Auxiliary Rotation Loss (CVPR 2019 · similarity 0.88)
- Unlocking Pre-trained Image Backbones for Semantic Image Synthesis (CVPR 2024 · similarity 0.88)
- High-Fidelity Image Generation With Fewer Labels (ICML 2019 · similarity 0.88)
- Learning to Predict Layout-to-image Conditional Convolutions for Semantic Image Synthesis (NeurIPS 2019 · 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.
