Browse Source

update readme.

master
wogong 5 years ago
parent
commit
3aa8b9b479
  1. 15
      README.md

15
README.md

@ -14,12 +14,15 @@ A PyTorch implementation for paper *[Unsupervised Domain Adaptation by Backpropa
## Note
- `Config()` 为针对特定任务的配置参数
- `MNISTmodel()` 完全按照论文中的结构,但是 feature 部分添加了 `Dropout2d()`,实验发现是否添加 `Dropout2d()` 对于最后的性能影响很大。最后实验重现结果高于论文,因为使用了额外的技巧,这里还有值得探究的地方。
- `SVHNmodel()` 无法理解论文中提出的结构,为自定义结构。最后实验重现结果完美。
- MNIST-MNISTM: `python mnist_mnistm.py`
- SVHN-MNIST: `python svhn_mnist.py`
- Amazon-Webcam: `python office.py` 由于预训练网络的问题,无法复现
- `MNISTmodel()`
- 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
- better result than paper
- `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
- `AlexModel`
- not successful, mainly due to the preprain model difference
## Result

Loading…
Cancel
Save