🙊 damped

Domain Adaptation Module for Privacy Enable and Distributed learning

damped is a library to experiment with hybrid deep learning architecture. It features a framework to incorporate off-tasks domain classifiers onto existing and well-defined toolkit easily. The goal is to have a minimal footprint on the main task network architecture, training loop, data preparation, and resource consumption.

🔗 References

List of the original papers that are implemented here:

1. Ganin, Y. et al. Domain-Adversarial Training of Neural Networks. arXiv:1505.07818 [cs, stat] (2015).
2. Anonymous. Multi-Step Decentralized Domain Adaptation. Paper under double-blind review for ICLR 2020 (2019).
3. Osia, S. A. et al. A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics. arXiv:1703.02952 [cs] (2017).
4. Srivastava, B., Bellet, A., Tommasi, M. & Vincent, E. Privacy-Preserving Adversarial Representation Learning in ASR: Reality or Illusion? Interspeech (2019) doi:10.21437/Interspeech.2019-2415.

Part of the code in this repository is inspired or borrowed from other implementations, especially:

• https://github.com/fungtion/DANN
• https://github.com/domainadaptation/salad
• https://github.com/espnet/espnet