Joint class generation on CelebA-ABMNSW (test)
91.37AccuracyAC-GAN
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| AC-GANOracle=true, Backbone=BigGAN2022.06 | 91.37 | 13.46 | 2.3 | 80 | 36 | 1.17 | 82 | |
| cGAN-PD + S2M samplingBackbone=BigGAN, MCMC iterations=10002022.06 | 79.74 | 10.3 | 2.45 | 85 | 47 | 1.45 | 88 | |
| CP-GANBackbone=BigGAN2022.06 | 79.09 | 14.97 | 2.45 | 74 | 46 | 0.95 | 77 | |
| GAN + S2M samplingBackbone=BigGAN, MCMC iterations=10002022.06 | 70.04 | 9.77 | 2.62 | 81 | 47 | 1.28 | 87 | |
| cGAN-PDOracle=true, Backbone=BigGAN2022.06 | 69.61 | 13.54 | 2.44 | 79 | 42 | 1.15 | 84 | |
| AC-GAN*Oracle=false, Backbone=BigGAN2022.06 | 23.97 | 11.92 | 2.44 | 77 | 45 | 1.11 | 82 | |
| cGAN-PD*Oracle=false, Backbone=BigGAN2022.06 | 15.28 | 11.09 | 2.32 | 80 | 47 | 1.19 | 85 |