bit-da/sct

This is a pytorch implementation of SCT (Semantic Correlation Transfer for Heterogeneous Domain Adaptation).

Prerequisites

- Python 3.6
- Pytorch 1.3.1
- numpy
- scipy
- matplotlib
- scikit_learn
- CUDA >= 8.0

Training Train and Evaluate

Using task amazon_surf to webcam_decaf as example:

python main.py --nepoch 5000 --d_common 256  --combine_pred Cosine_threshold  --cuda 3 --source amazon_surf --target webcam_decaf --lr_first 0.05 --lr 0.006 --mean_loss 1 --shift_loss 0.05 --cos_loss 0.1 --nepoch_first 1500

Contact

If you have any problem about our code, feel free to contact

• [email protected]
• [email protected]
• [email protected]

or describe your problem in Issues.

sct's Issues

关于数据集的问题

老师和同学们好，我对这个论文也比较感兴趣，想问问这个论文对应的数据集可以提供一下吗？谢谢各位

ModuleNotFoundError: No module named 'kernel_kmeans'

运行发现ModuleNotFoundError: No module named 'kernel_kmeans'，貌似没有这个包可以安装