|
|
@ -21,17 +21,19 @@ A PyTorch implementation for paper *[Unsupervised Domain Adaptation by Backpropa |
|
|
|
- `SVHNmodel()` |
|
|
|
- network structure proposed in the paper may be wrong for both 32x32 and 28x28 inputs |
|
|
|
- change last conv layer's filter to 4x4, get similar(actually higher) result |
|
|
|
- `GTSRBmodel()` |
|
|
|
- not successful |
|
|
|
- `AlexModel` |
|
|
|
- not successful, mainly due to the preprain model difference |
|
|
|
|
|
|
|
## Result |
|
|
|
|
|
|
|
| | MNIST-MNISTM | SVHN-MNIST | Amazon-Webcam |Amazon-Webcam10 | |
|
|
|
| :------------------: | :------------: | :--------: | :-----------: |:-------------: | |
|
|
|
| Source Only | 0.5225 | 0.5490 | 0.6420 | 0. | |
|
|
|
| DANN(paper) | 0.7666 | 0.7385 | 0.7300 | 0. | |
|
|
|
| This Repo Source Only| - | - | - | 0. | |
|
|
|
| This Repo | 0.8400 | 0.7339 | 0.6528 | 0. | |
|
|
|
| | MNIST-MNISTM | SVHN-MNIST | Amazon-Webcam | SYNDIGITS-SVHN | SYNSIGNS-GTSRB | |
|
|
|
| :------------------: | :------------: | :--------: | :-----------: |:-------------: |:-------------: | |
|
|
|
| Source Only | 0.5225 | 0.5490 | 0.6420 | 0. | 0. | |
|
|
|
| DANN(paper) | 0.7666 | 0.7385 | 0.7300 | 0.9109 | 0.7900 | |
|
|
|
| This Repo Source Only| - | - | - | 0. | 0. | |
|
|
|
| This Repo | 0.8400 | 0.7339 | 0.6528 | 0.8200 | 0.6200 | |
|
|
|
|
|
|
|
## Credit |
|
|
|
|
|
|
|