FADA-Pytorch

pytorch implement for the NIPS 2017 paper Few-Shot Adversarial Domain Adaptation

Install dependencies

• This code has been tested on Python 3.6. pytorch 0.4.
• Install PyTorch, torchvision and SciPy: pip install -r requirements.txt

Set up the dataset

• Run python main.pyand get MNIST and SVHN dataset in ./data/

Train the model

• Run python main.py.
  • Step 1 refers to pretrain model using MNIST dataset
  • Setp 2 refers to pretrain DCD using G1,G2,G3,G4.
  • Step 3 refers to train g,h and DCD iteratively.

Evaluate

• Using 7 shot in target domain , training step 3 after 30 epoches ,I got the same result 47% acc as shown in papers