Research map / Image generation / GANs
Continuous Conditional Generative Adversarial Networks: Novel Empirical Losses and Label Input Mechanisms
Xin Ding, Yongwei Wang, Zuheng Xu, William J. Welch, Z. Jane Wang
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 |
|---|---|---|---|
| use improved label input (ILI) instead of naive label input (NLI) (ablation) | CcGAN (SVDL) · Cell-200 (64x64) | Intra-FID: 56.078 -> 7.266 | Improved |
| use ILI instead of NLI (ablation) | CcGAN (SVDL) · RC-49 (64x64) | Intra-FID: 0.515 -> 0.389 | Improved |
| use ILI instead of NLI (ablation) | CcGAN (SVDL) · UTKFace (64x64) | Intra-FID: 0.547 -> 0.425 | Improved |
| use ILI instead of NLI (ablation) | CcGAN (SVDL) · Steering Angle (64x64) | Intra-FID: 1.866 -> 1.546 | Improved |
| use soft vicinal loss (SVDL) instead of hard vicinal loss (HVDL) (ablation) | CcGAN (ILI) · RC-49 (64x64) | Intra-FID: 0.424 -> 0.389 | Improved |
| use SVDL instead of HVDL (ablation) | CcGAN (ILI) · Cell-200 (64x64) | Intra-FID: 8.759 -> 7.266 | Improved |
| use SVDL instead of HVDL (ablation) | CcGAN (NLI) · Cell-200 (64x64) | Intra-FID: 50.052 -> 56.078 | Hurt |
| replace class binning (cGAN) with CcGAN vicinal losses and ILI | cGAN (150 classes) · RC-49 (64x64) | Intra-FID: 1.720 -> 0.389 | Improved |
| use CcGAN (SVDL+ILI) instead of cGAN (concat) | cGAN (concat) · RC-49 (64x64) | Label Score: 30.212 -> 1.940 | Improved |
| use SFID with nearby real images instead of Intra-FID on seen angles | evaluation metric · RC-49 HVDL+ILI | FID score: 1.032 -> 0.401 | Improved |
| add DiffAugment and BigGAN to CcGAN (SVDL+ILI) versus SNGAN | CcGAN (SVDL+ILI) · RC-49 | Intra-FID: 0.389 -> 0.086 | Improved |
| use CcGAN (SVDL+ILI) with SNGAN instead of ReACGAN with advanced training | ReACGAN · RC-49 | Intra-FID: 0.178 -> 0.389 | Hurt |
Closest papers by meaning
- CcGAN: Continuous Conditional Generative Adversarial Networks for Image Generation (ICLR 2021 · similarity 0.98)
- Dual Projection Generative Adversarial Networks for Conditional Image Generation (ICCV 2021 · similarity 0.93)
- Robust Conditional Generative Adversarial Networks (ICLR 2019 · similarity 0.92)
- Turning Waste into Wealth: Leveraging Low-Quality Samples for Enhancing Continuous Conditional Generative Adversarial Networks (AAAI 2024 · similarity 0.92)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.89)
- Fusing Conditional Submodular GAN and Programmatic Weak Supervision (AAAI 2024 · similarity 0.89)
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.
