Digit Classification on USPS -> MNIST
99.1AccuracyTarget Only
Evaluation results
| Method | Links | |
|---|---|---|
| Target OnlyEvaluation Protocol=Target Only2021.03 | 99.1 | |
| LC + CycleGAN2021.03 | 98.3 | |
| DRANetDirectionality=Bi-directional2021.03 | 97.8 | |
| distributionally robust learning frameworkBase classifier=STAR, Hyperparameter selection=Validation-based, Target sample size=10^22026.01 | 97.3 | |
| DRANetDirectionality=Tri-directional2021.03 | 96.9 | |
| distributionally robust learning frameworkBase classifier=STAR, Hyperparameter selection=LODO-CV, Target sample size=10^22026.01 | 96.6 | |
| CyCADA2021.03 | 96.5 | |
| MCDTarget sample size=10^22026.01 | 96.1 | |
| ADDA2021.03 | 95.2 | |
| STARTarget sample size=10^22026.01 | 94.5 | |
| distributionally robust learning frameworkBase classifier=STAR, Hyperparameter selection=Validation-based, Target sample size=102026.01 | 93 | |
| ATDOCTarget sample size=10^22026.01 | 92.6 | |
| distributionally robust learning frameworkBase classifier=CDAN, Hyperparameter selection=Validation-based, Target sample size=10^22026.01 | 91.5 | |
| Generate To AdaptBackbone=Modified LeNet2017.04 | 90.8 | |
| distributionally robust learning frameworkBase classifier=ERM, Hyperparameter selection=Validation-based, Target sample size=10^22026.01 | 90.3 | |
| ADDABackbone=Modified LeNet2017.04 | 90.1 | |
| CoGANBackbone=Modified LeNet2017.04 | 89.1 | |
| CoGAN2021.03 | 89.1 | |
| distributionally robust learning frameworkBase classifier=STAR, Hyperparameter selection=LODO-CV, Target sample size=102026.01 | 87.6 | |
| ATDOCTarget sample size=102026.01 | 87 | |
| distributionally robust learning frameworkBase classifier=CDAN, Hyperparameter selection=Validation-based, Target sample size=102026.01 | 87 | |
| distributionally robust learning frameworkBase classifier=ERM, Hyperparameter selection=Validation-based, Target sample size=102026.01 | 86.4 | |
| MCDTarget sample size=102026.01 | 85.9 | |
| STARTarget sample size=102026.01 | 85.2 | |
| CORALTarget sample size=10^22026.01 | 75.9 | |
| DANNTarget sample size=10^22026.01 | 74.8 | |
| DRCNBackbone=Modified LeNet2017.04 | 73.7 | |
| RevGradBackbone=Modified LeNet2017.04 | 73 | |
| DANN2021.03 | 73 | |
| CORALTarget sample size=102026.01 | 64.7 | |
| ERM (Src-only)Target sample size=None (Source-only)2026.01 | 60.4 | |
| CDANTarget sample size=10^22026.01 | 58.3 | |
| MK-MMDTarget sample size=10^22026.01 | 57.6 | |
| Source onlyBackbone=Modified LeNet2017.04 | 57.1 | |
| CDANTarget sample size=102026.01 | 54.8 | |
| DANNTarget sample size=102026.01 | 51.1 | |
| Source OnlyEvaluation Protocol=Source Only2021.03 | 44.9 | |
| MK-MMDTarget sample size=102026.01 | 32.1 |