Tensorflow codes for ICML2018, Learning Semantic Representations for Unsupervised Domain Adaptation
Based on adversarial adaptation, we propose a Pseudo Centroid Alignment Objective
to enforce Semantic Transfer
. If you are limited to use a relative-small batch size (64 for 31-classification or 100 for 1000-classification), you might be interested in our Moving Centroid Alignment
.
If you find this useful for your research, we would be appreciated if you cite the following papers:
@inproceedings{xie2018learning,
title={Learning Semantic Representations for Unsupervised Domain Adaptation},
author={Xie, Shaoan and Zheng, Zibin and Chen, Liang and Chen, Chuan},
booktitle={International Conference on Machine Learning},
pages={5419--5428},
year={2018}
}
My work is based on DANN. During my reimplementation of DANN, I noticed following problems worth attention for reproduing DANN and our work MSTN. Hope these could help you. :)
If you have any problem about this library, please create an Issue or send us an Email at:
For digits dataset, the code is modified from Here. For real-world dataset, the code is based on Here.