Digit Classification on SVHN to MNIST (test)
99.4AccuracyVMT
Evaluation results
| Method | Links | |
|---|---|---|
| VMTbackbone=Lipschitz-constraint-based (larger)2019.05 | 99.4 | |
| DIRTbackbone=Lipschitz-constraint-based (larger)2019.05 | 99.4 | |
| SRDA-Gnoise_method=Gaussian, image_level_optimization=false2019.05 | 99.17 | |
| SRDA-Vnoise_method=VAT, image_level_optimization=false2019.05 | 98.9 | |
| MCD2019.05 | 96.2 | |
| SRDA-Fnoise_method=FGSM, image_level_optimization=false2019.05 | 95.96 | |
| ASSC2019.05 | 95.7 | |
| VADAbackbone=Lipschitz-constraint-based (larger)2019.05 | 94.5 | |
| dkdHTLNumber of Target Data per Class=72021.06 | 90.3 | |
| TOHANNumber of Target Data per Class=72021.06 | 89.7 | |
| dkdHTLNumber of Target Data per Class=62021.06 | 89.6 | |
| SRDA-G*noise_method=Gaussian, image_level_optimization=true2019.05 | 89.51 | |
| dkdHTLNumber of Target Data per Class=52021.06 | 89.5 | |
| SRDA-V*noise_method=VAT, image_level_optimization=true2019.05 | 89.47 | |
| dkdHTLNumber of Target Data per Class=32021.06 | 88.5 | |
| dkdHTLNumber of Target Data per Class=42021.06 | 88.2 | |
| TOHANNumber of Target Data per Class=62021.06 | 88 | |
| TOHANNumber of Target Data per Class=52021.06 | 87.1 | |
| TOHANNumber of Target Data per Class=42021.06 | 86.5 | |
| ATDA2019.05 | 86.2 | |
| TOHANNumber of Target Data per Class=32021.06 | 84.2 | |
| dkdHTLNumber of Target Data per Class=22021.06 | 83.4 | |
| TOHANNumber of Target Data per Class=22021.06 | 83.3 | |
| DSN2019.05 | 82.7 | |
| Source Only (Lipschitz config)adaptation_protocol=No adaptation, backbone=Lipschitz-constraint-based (larger)2019.05 | 82.4 | |
| DRCN2019.05 | 82 | |
| TOHANNumber of Target Data per Class=12021.06 | 76 | |
| ADDA2019.05 | 76 | |
| dkdHTLNumber of Target Data per Class=12021.06 | 71.2 | |
| DAN2019.05 | 71.1 | |
| DANN2019.05 | 71.1 | |
| Source Onlyadaptation_protocol=No adaptation2019.05 | 67.1 | |
| SRDA-F*noise_method=FGSM, image_level_optimization=true2019.05 | 22.7 |