Spherical Space Domain Adaptation with Robust Pseudo-Label Loss
Abstract
Adversarial domain adaptation (DA) has been an effective approach for learning domain-invariant features by adversarial training. In this paper, we propose a novel adversarial DA approach completely defined in spherical feature space, in which we define spherical classifier for label prediction and spherical domain discriminator for discriminating domain labels. To utilize pseudo-label robustly, we develop a robust pseudo-label loss in the spherical feature space, which weights the importance of estimated labels of target data by posterior probability of correct labeling, modeled by Gaussian-uniform mixture model in spherical feature space. Extensive experiments show that our method achieves state-of-the-art results, and also confirm effectiveness of spherical classifier, spherical discriminator and spherical robust pseudo-label loss.
Cite
Text
Gu et al. "Spherical Space Domain Adaptation with Robust Pseudo-Label Loss." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020. doi:10.1109/CVPR42600.2020.00912Markdown
[Gu et al. "Spherical Space Domain Adaptation with Robust Pseudo-Label Loss." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020.](https://mlanthology.org/cvpr/2020/gu2020cvpr-spherical/) doi:10.1109/CVPR42600.2020.00912BibTeX
@inproceedings{gu2020cvpr-spherical,
title = {{Spherical Space Domain Adaptation with Robust Pseudo-Label Loss}},
author = {Gu, Xiang and Sun, Jian and Xu, Zongben},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year = {2020},
doi = {10.1109/CVPR42600.2020.00912},
url = {https://mlanthology.org/cvpr/2020/gu2020cvpr-spherical/}
}