Bin-Cao/TrAdaboost

The transfer learning code for understanding and teaching : Boosting for transfer learning with single / multiple source(s)

🤝🤝🤝 Please star ⭐️ it for promoting open source projects 🌍 ! Thanks !

if you have any questions or need help, you are welcome to contact me

If you are using this code, please cite:

• Cao Bin, Zhang Tong-yi, Xiong Jie, Zhang Qian, Sun Sheng. Package of Boosting-based transfer learning [2023SR0525555], 2023, Software copyright, GitHub : github.com/Bin-Cao/TrAdaboost.

Star History

TrAdaBoost : Boosting for transfer learning

Transfer learning allows leveraging the knowledge of source domains, available a priori, to help training a classifier for a target domain, where the available data is scarce.

Models

(1) classification

• TrAdaBoost
• MultiSourceTrAdaBoost
• TaskTrAdaBoost
• ExpBoost
• Improved ExpBoost

(2) Regression

• Transfer Stacking
• TrAdaboost R2
• Two stage TrAdaboost R2
• Two stage TrAdaboost R2 rv

Written using Python, which is suitable for operating systems, e.g., Windows/Linux/MAC OS etc.

中文介绍(陆续更新）

• TrAdaBoost

Note

author_email='bcao@shu.edu.com'
maintainer='CaoBin'
maintainer_email='bcao@shu.edu.cn' 
license='MIT License'
url='https://github.com/Bin-Cao/TrAdaboost'
python_requires='>=3.7'

References

.. [1] Dai, W., Yang, Q., et al. (2007). Boosting for Transfer Learning.(2007), 193--200. In Proceedings of the 24th international conference on Machine learning.

.. [2] Yao, Y., & Doretto, G. (2010, June) Boosting for transfer learning with multiple sources. IEEE. DOI: 10.1109/CVPR.2010.5539857

.. [3] Rettinger, A., Zinkevich, M., & Bowling, M. (2006, July). Boosting expert ensembles for rapid concept recall. In Proceedings of the National Conference on Artificial Intelligence (Vol. 21, No. 1, p. 464). Menlo Park, CA; Cambridge, MA; London; AAAI Press; MIT Press; 1999.

.. [4] Pardoe, D., & Stone, P. (2010, June). Boosting for regression transfer. In Proceedings of the 27th International Conference on International Conference on Machine Learning (pp. 863-870).

About

Maintained by Bin Cao. Please feel free to open issues in the Github or contact Bin Cao (bcao686@connect.hkust-gz.edu.cn) in case of any problems/comments/suggestions in using the code.

---

Transfer learning links

1 : Instance-based transfer learning

• Instance selection (marginal distributions are same while conditional distributions are different) :

  TrAdaboost

• Instance re-weighting (conditional distributions are same while marginal distributions are different) :

  KMM

2 : Feature-based transfer learning

• Explicit distance:

  • case 1 : marginal distributions are same while conditional distributions are different:

    TCA(MMD based) ; DAN(MK-MMD based)

  • case 1 : conditional distributions are same while marginal distributions are different

    JDA

  • case 3 : Both marginal distributions and conditional distributions are different

    DDA

• Implicit distance :

  DANN

3 : Parameter-based transfer learning

• Pretraining + fine tune

---