Home

Awesome

SSHT

"SSHT" (Semi-supervised Source Hypothesis Transfer.)
Paper: Learning Invariant Representation with Consistency and Diversity for Semi-supervised Source Hypothesis Transfer

Environment

If you use conda, just run the following:

conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=10.0

Dataset

Office-31 can be found here.
Office-Home can be found here.
DomainNet can be found here.
Visda-2017 can be found here.
Here, we provide the label lists for the above datasets, for UDA, SSDA (1 shot and 3 shot).

Train and Test

Please refer run_sh.md.

Citation

If you use this code for your research, please consider citing:

@article{wang2021learning,
  title={Learning Invariant Representation with Consistency and Diversity for Semi-supervised Source Hypothesis Transfer},
  author={Wang, Xiaodong and Zhuo, Junbao and Cui, Shuhao and Wang, Shuhui},
  journal={arXiv preprint arXiv:2107.03008},
  year={2021}
}

Contact

If you have any question, contact to me: