|
@ -14,6 +14,8 @@ A PyTorch implementation for paper *[Unsupervised Domain Adaptation by Backpropa |
|
|
|
|
|
|
|
|
## Note |
|
|
## Note |
|
|
|
|
|
|
|
|
|
|
|
Before running the training code, make sure that `DATASETDIR` is set to dataset directory. |
|
|
|
|
|
|
|
|
- `MNISTmodel()` |
|
|
- `MNISTmodel()` |
|
|
- basically the same network structure as proposed in the paper, expect for adding dropout layer in feature extractor |
|
|
- basically the same network structure as proposed in the paper, expect for adding dropout layer in feature extractor |
|
|
- large gap exsits between with and w/o dropout layer |
|
|
- large gap exsits between with and w/o dropout layer |
|
|