Generative Modeling on MNIST (test)
99.32N-ELBOSTGS
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| STGSLatent Structure (Variables x Categories)=8x16, Hyper-parameter selection=Based on Train -ELBO2023.04 | 99.32 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=8x16, Hyper-parameter selection=Based on Train -ELBO2023.04 | 100.63 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=10x30, Hyper-parameter selection=Based on Train -ELBO2023.04 | 100.75 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=10x30, Hyper-parameter selection=Based on Train -ELBO2023.04 | 100.78 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=16x12, Hyper-parameter selection=Based on Train -ELBO2023.04 | 100.85 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=16x12, Hyper-parameter selection=Based on Train -ELBO2023.04 | 101.28 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=8x16, Hyper-parameter selection=Based on Train -ELBO2023.04 | 101.44 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=10x30, Hyper-parameter selection=Based on Train -ELBO2023.04 | 101.61 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=8x16, Hyper-parameter selection=Based on Train -ELBO2023.04 | 102.12 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=4x24, Hyper-parameter selection=Based on Train -ELBO2023.04 | 102.4 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=16x12, Hyper-parameter selection=Based on Train -ELBO2023.04 | 102.49 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=4x24, Hyper-parameter selection=Based on Train -ELBO2023.04 | 102.76 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=64x8, Hyper-parameter selection=Based on Train -ELBO2023.04 | 102.91 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=4x24, Hyper-parameter selection=Based on Train -ELBO2023.04 | 103.6 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=10x30, Hyper-parameter selection=Based on Train -ELBO2023.04 | 103.62 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=4x24, Hyper-parameter selection=Based on Train -ELBO2023.04 | 103.95 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=16x12, Hyper-parameter selection=Based on Train -ELBO2023.04 | 104.23 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=64x8, Hyper-parameter selection=Based on Train -ELBO2023.04 | 105.44 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=AVG, Hyper-parameter selection=Based on Train -ELBO2023.04 | 105.74 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=64x8, Hyper-parameter selection=Based on Train -ELBO2023.04 | 106.2 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=AVG, Hyper-parameter selection=Based on Train -ELBO2023.04 | 106.85 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=AVG, Hyper-parameter selection=Based on Train -ELBO2023.04 | 106.89 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=AVG, Hyper-parameter selection=Based on Train -ELBO2023.04 | 109.03 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=4x24, Hyper-parameter selection=Based on Train -ELBO2023.04 | 113.41 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=64x8, Hyper-parameter selection=Based on Train -ELBO2023.04 | 113.54 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=8x16, Hyper-parameter selection=Based on Train -ELBO2023.04 | 114.25 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=16x12, Hyper-parameter selection=Based on Train -ELBO2023.04 | 114.48 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=64x8, Hyper-parameter selection=Based on Train -ELBO2023.04 | 115.43 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=10x30, Hyper-parameter selection=Based on Train -ELBO2023.04 | 118.46 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=AVG, Hyper-parameter selection=Based on Train -ELBO2023.04 | 118.85 | — | — | — | — | — | — | — | — | — | — | |
| ReinMaxLatent Structure (Variables x Categories)=8x4, Hyper-parameter selection=Based on Train -ELBO2023.04 | 126.89 | — | — | — | — | — | — | — | — | — | — | |
| GR-MCKLatent Structure (Variables x Categories)=8x4, Hyper-parameter selection=Based on Train -ELBO2023.04 | 127.9 | — | — | — | — | — | — | — | — | — | — | |
| STGSLatent Structure (Variables x Categories)=8x4, Hyper-parameter selection=Based on Train -ELBO2023.04 | 128.09 | — | — | — | — | — | — | — | — | — | — | |
| GST-1.0Latent Structure (Variables x Categories)=8x4, Hyper-parameter selection=Based on Train -ELBO2023.04 | 128.2 | — | — | — | — | — | — | — | — | — | — | |
| STLatent Structure (Variables x Categories)=8x4, Hyper-parameter selection=Based on Train -ELBO2023.04 | 137.06 | — | — | — | — | — | — | — | — | — | — | |
| AdaptiveTemperature Strategy=Adaptive2026.03 | — | — | -684.56 | — | — | — | — | — | 0.0161 | 310.97 | — | |
| Attentive VAEL=152024.12 | — | — | — | — | — | — | — | 77.63 | — | — | — | |
| BEGAN2019.05 | — | 13.1 | — | — | — | — | — | — | — | — | — | |
| BIVAL=62024.12 | — | — | — | — | — | — | — | 78.41 | — | — | — | |
| CatGANunsupervised=true2015.11 | — | — | 2,376 | — | — | — | — | — | — | — | — | |
| CR-NVAEL=15, Data augmentation=true2024.12 | — | — | — | — | — | — | — | 76.93 | — | — | — | |
| DRAGAN2019.05 | — | 7.6 | — | — | — | — | — | — | — | — | — | |
| DVP-VAEL=8, Averaged over 4 random seeds=true2024.12 | — | — | — | — | — | — | — | 77.1 | — | — | — | |
| FFJORDArchitecture=VAE2024.03 | — | — | — | — | 55.9 | 17.3 | 96.4 | — | — | — | — | |
| FixedTemperature Strategy=T=12026.03 | — | — | -714.29 | — | — | — | — | — | 0.0168 | 65.23 | — | |
| FixedTemperature Strategy=T=T*2026.03 | — | — | -689.39 | — | — | — | — | — | 0.0161 | 65.82 | — | |
| GAN2015.11 | — | — | 2,252 | — | — | — | — | — | — | — | — | |
| GLANNperceptual loss=true2019.05 | — | 8.6 | — | — | — | — | — | — | — | — | — | |
| GLANNperceptual loss=true2019.05 | — | 8.6 | — | — | — | — | — | — | — | — | — | |
| GLF2019.05 | — | 8.2 | — | — | — | — | — | — | — | — | — | |
| GLFperceptual loss=true2019.05 | — | 5.8 | — | — | — | — | — | — | — | — | — | |
| GLF2019.05 | — | 8.2 | — | — | — | — | — | — | — | — | — | |
| GLFperceptual loss=true2019.05 | — | 5.8 | — | — | — | — | — | — | — | — | — | |
| GMMN2015.11 | — | — | 1,472 | — | — | — | — | — | — | — | — | |
| GMMN + AEinput_type=hidden layer activations of an AE2015.11 | — | — | 2,822 | — | — | — | — | — | — | — | — | |
| GSN2015.11 | — | — | 2,141 | — | — | — | — | — | — | — | — | |
| IA-VAENumber of random seeds=102026.04 | — | — | — | -78.68 | — | — | — | — | — | — | 0.01 | |
| IAF-VAE2024.12 | — | — | — | — | — | — | — | 79.1 | — | — | — | |
| IWK=12021.07 | — | — | — | -92.4 | — | — | — | — | — | — | — | |
| IWK=82021.07 | — | — | — | -89.9 | — | — | — | — | — | — | — | |
| IWK=162021.07 | — | — | — | -89.3 | — | — | — | — | — | — | — | |
| IWK=322021.07 | — | — | — | -88.8 | — | — | — | — | — | — | — | |
| IWK=642021.07 | — | — | — | -88.5 | — | — | — | — | — | — | — | |
| IWK=1, Architecture=1 hidden layer2021.07 | — | — | — | -93.4 | — | — | — | — | — | — | — | |
| IWK=8, Architecture=1 hidden layer2021.07 | — | — | — | -90.5 | — | — | — | — | — | — | — | |
| IWK=16, Architecture=1 hidden layer2021.07 | — | — | — | -89.9 | — | — | — | — | — | — | — | |
| IWK=32, Architecture=1 hidden layer2021.07 | — | — | — | -89.4 | — | — | — | — | — | — | — | |
| IWK=64, Architecture=1 hidden layer2021.07 | — | — | — | -89 | — | — | — | — | — | — | — | |
| LSGAN2019.05 | — | 7.8 | — | — | — | — | — | — | — | — | — | |
| LVAEL=52024.12 | — | — | — | — | — | — | — | 81.74 | — | — | — | |
| MM GAN2019.05 | — | 9.8 | — | — | — | — | — | — | — | — | — | |
| NS GAN2019.05 | — | 6.8 | — | — | — | — | — | — | — | — | — | |
| NVAEL=152024.12 | — | — | — | — | — | — | — | 78.01 | — | — | — | |
| OU-VAEL=52024.12 | — | — | — | — | — | — | — | 81.1 | — | — | — | |
| PaddingFlowArchitecture=VAE, Base Method=FFJORD, Noise=Padding-dimensional2024.03 | — | — | — | — | 36.1 | 11 | 100 | — | — | — | — | |
| PPMMEstimation Method=PPMM, Latent Dimension=82021.06 | — | 0.17 | — | — | — | — | — | — | — | — | — | |
| RAE + GMM2019.05 | — | 10.8 | — | — | — | — | — | — | — | — | — | |
| RAE + GMM2019.05 | — | 10.8 | — | — | — | — | — | — | — | — | — | |
| RANDOMEstimation Method=RANDOM, Latent Dimension=82021.06 | — | 4.62 | — | — | — | — | — | — | — | — | — | |
| SLICEDEstimation Method=SLICED, Number of projections (k)=10, Latent Dimension=82021.06 | — | 2.98 | — | — | — | — | — | — | — | — | — | |
| SLICEDEstimation Method=SLICED, Number of projections (k)=20, Latent Dimension=82021.06 | — | 3.04 | — | — | — | — | — | — | — | — | — | |
| SLICEDEstimation Method=SLICED, Number of projections (k)=50, Latent Dimension=82021.06 | — | 3.12 | — | — | — | — | — | — | — | — | — | |
| Two-Stage VAE2019.05 | — | 10.9 | — | — | — | — | — | — | — | — | — | |
| Two-Stage VAE2019.05 | — | 10.9 | — | — | — | — | — | — | — | — | — | |
| UHAK=12021.07 | — | — | — | -92.4 | — | — | — | — | — | — | — | |
| UHAK=82021.07 | — | — | — | -89.2 | — | — | — | — | — | — | — | |
| UHAK=162021.07 | — | — | — | -88.5 | — | — | — | — | — | — | — | |
| UHAK=322021.07 | — | — | — | -88.1 | — | — | — | — | — | — | — | |
| UHAK=642021.07 | — | — | — | -87.1 | — | — | — | — | — | — | — | |
| UHAK=1, Architecture=1 hidden layer2021.07 | — | — | — | -93.4 | — | — | — | — | — | — | — | |
| UHAK=8, Architecture=1 hidden layer2021.07 | — | — | — | -89.8 | — | — | — | — | — | — | — | |
| UHAK=16, Architecture=1 hidden layer2021.07 | — | — | — | -88.8 | — | — | — | — | — | — | — | |
| UHAK=32, Architecture=1 hidden layer2021.07 | — | — | — | -88.1 | — | — | — | — | — | — | — | |
| UHAK=64, Architecture=1 hidden layer2021.07 | — | — | — | -87.6 | — | — | — | — | — | — | — | |
| VAE2019.05 | — | 28.2 | — | — | — | — | — | — | — | — | — | |
| VAEprior=flow2019.05 | — | 28.3 | — | — | — | — | — | — | — | — | — | |
| VAEposterior=flow2019.05 | — | 26.7 | — | — | — | — | — | — | — | — | — | |
| VAE2019.05 | — | 28.2 | — | — | — | — | — | — | — | — | — | |
| VAENumber of random seeds=102026.04 | — | — | — | -79.42 | — | — | — | — | — | — | — | |
| VAE+flow posterior2019.05 | — | 26.7 | — | — | — | — | — | — | — | — | — |