Image Generation on Fashion-MNIST
1.49FIDReal
Evaluation results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Real2026.01 | 1.49 | — | — | — | 8.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SIGCategory=iterative, T=12020.07 | 13.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PrivImageepsilon=12026.01 | 16.1 | — | — | 79.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PDP-Diffusionepsilon=12026.01 | 16.6 | — | — | 79.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| WGANCategory=adversarial training2020.07 | 21.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DuelGANreported_in_original_paper=false2021.01 | 21.73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PDP-Diffusion-Proepsilon=12026.01 | 23.4 | — | — | 81.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| WGAN GPCategory=adversarial training2020.07 | 24.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DCGANreported_in_original_paper=false2021.01 | 24.78 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PrivImage-Proepsilon=12026.01 | 25.4 | — | — | 78.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FETA-Proepsilon=12026.01 | 27.8 | — | — | 83.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PAECategory=AE based2020.07 | 28 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| WGANreported_in_original_paper=true2021.01 | 28.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| two-stage VAECategory=AE based2020.07 | 29.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| D2GANreported_in_original_paper=false2021.01 | 29.33 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Best default GANCategory=adversarial training2020.07 | 32 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LSGANreported_in_original_paper=false2021.01 | 43 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PEepsilon=12026.01 | 48.8 | — | — | 47.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=MNIST2023.05 | 56.59 | 0.0387 | 18.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CWAECategory=AE based2020.07 | 57.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DCGANInput Distribution=Non-Parametric (R^100)2023.05 | 62.42 | 0.0426 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DRAGANreported_in_original_paper=false2021.01 | 62.64 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DCGANInput Distribution=Gamma (R^100)2023.05 | 65.36 | 0.0513 | 19.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=Ukiyo-E2023.05 | 66.9 | 0.0475 | 23.29 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWAECategory=AE based, Reference=Kolouri et al., 20182020.07 | 74.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DCGANInput Distribution=Gaussian (R^100)2023.05 | 76.6 | 0.0537 | 22.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=SVHN2023.05 | 79.14 | 0.0526 | 24.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=CelebA2023.05 | 81.38 | 0.0604 | 24.73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Oursprivacy_budget_epsilon=10, privacy_budget_delta=1e-5, sigma_noise=0.1, gamma_recon=12026.01 | 83.48 | — | — | — | 6.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DP-GAN-DPACprivacy_budget_epsilon=10, privacy_budget_delta=1e-52026.01 | 90.77 | — | — | — | 6.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=CIFAR-102023.05 | 92.6 | 0.0658 | 30.21 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=LSUN Churches2023.05 | 102.9 | 0.0774 | 33.87 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DCGANInput Distribution=Gaussian (R^HxWxC)2023.05 | 119.2 | 0.0905 | 28.96 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPSinkhornprivacy_budget_epsilon=10, privacy_budget_delta=1e-52026.01 | 129.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spider DCGANInput Distribution=TinyImageNet2023.05 | 130.5 | 0.0883 | 22.26 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GS-WGANprivacy_budget_epsilon=10, privacy_budget_delta=1e-52026.01 | 131.34 | — | — | — | 5.32 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWFCategory=iterative2020.07 | 207.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Flow-GAN (ADV)Category=adversarial training2020.07 | 216.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| C-VAE2025.08 | — | 0.1835 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 195.3205 | 193.0972 | |
| DP-GAN-DPACClassifier Architecture=MLP2026.01 | — | — | — | — | — | 74 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DP-GAN-DPACClassifier Architecture=CNN2026.01 | — | — | — | — | — | 71 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPSinkhornClassifier Architecture=MLP2026.01 | — | — | — | — | — | 73 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPSinkhornClassifier Architecture=CNN2026.01 | — | — | — | — | — | 71 | — | — | — | — | — | — | — | — | — | — | — | — | |
| EFSGDClassifier Architecture=MLP2026.01 | — | — | — | — | — | 74 | — | — | — | — | — | — | — | — | — | — | — | — | |
| EFSGDClassifier Architecture=CNN2026.01 | — | — | — | — | — | 70 | — | — | — | — | — | — | — | — | — | — | — | — | |
| G-VAE2025.08 | — | 0.1828 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 179.8126 | 179.2981 | |
| GS-WGANClassifier Architecture=MLP2026.01 | — | — | — | — | — | 65 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GS-WGANClassifier Architecture=CNN2026.01 | — | — | — | — | — | 65 | — | — | — | — | — | — | — | — | — | — | — | — | |
| L-VAE2025.08 | — | 0.1847 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 181.4542 | 179.5956 | |
| NegBio-VAEvariant=MC-G2025.08 | — | 0.1468 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 127.5248 | 125.9497 | |
| NegBio-VAEvariant=MC-C2025.08 | — | 0.1688 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 148.9795 | 147.7799 | |
| NegBio-VAEvariant=DS-G2025.08 | — | 0.1517 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 133.0601 | 132.8822 | |
| NegBio-VAEvariant=DS-C2025.08 | — | 0.1763 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 155.5468 | 154.1402 | |
| P-VAE2025.08 | — | 0.1667 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 145.9776 | 146.0128 | |
| PQWGAN2026.03 | — | — | — | — | — | — | 84.8 | 92.2 | 80.3 | 89 | 83.8 | 87.5 | 78.6 | 90.5 | 75.4 | 85.4 | — | — | |
| QINR-AE2026.03 | — | — | — | — | — | — | 92 | 92.2 | 92.5 | 91.8 | 94.2 | 75.5 | 91.8 | 89.8 | 90.3 | 93.1 | — | — | |
| QINR-QGAN2026.03 | — | — | — | — | — | — | 90.8 | 95.3 | 87.4 | 93.4 | 90.4 | 93 | 87.4 | 95.6 | 85.1 | 91.7 | — | — | |
| QINR-VAEmode=prior2026.03 | — | — | — | — | — | — | 85.6 | 87.4 | 87.9 | 85.2 | 88.4 | 57.6 | 84.7 | 81.8 | 79 | 82.8 | — | — | |
| QINR-VAEmode=recons2026.03 | — | — | — | — | — | — | 91.6 | 94.3 | 93.1 | 92.1 | 93.7 | 76.9 | 91.5 | 89.4 | 90.6 | 92.9 | — | — | |
| Quantum AnoGAN2026.03 | — | — | — | — | — | — | 90.8 | 95.4 | 87.8 | 93.5 | 90.4 | 93.1 | 87.4 | 95.7 | 85 | 91.6 | — | — |