Continual Domain Adaptation Through Pruning-Aided Domain-Specific Weight Modulation

Prasanna B, Sunandini Sanyal, R. Venkatesh Babu

CVPRW 2023 pp. 2457-2463

doi:10.1109/CVPRW59228.2023.00244 /cvprw/2023/b2023cvprw-continual/

Abstract

In this paper, we propose to develop a method to address unsupervised domain adaptation (UDA) in a practical setting of continual learning (CL). The goal is to update the model on continually changing domains while preserving domain-specific knowledge to prevent catastrophic forgetting of past-seen domains. To this end, we build a framework for preserving domain-specific features utilizing the inherent model capacity via pruning. We also perform effective inference using a novel batch-norm based metric to predict the final model parameters to be used accurately. Our approach achieves not only state-of-the-art performance but also prevents catastrophic forgetting of past domains significantly. Our code is made publicly available. 1.

PDF CVPRW Semantic Scholar

Cite

Text

B et al. "Continual Domain Adaptation Through Pruning-Aided Domain-Specific Weight Modulation." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2023. doi:10.1109/CVPRW59228.2023.00244

Markdown

[B et al. "Continual Domain Adaptation Through Pruning-Aided Domain-Specific Weight Modulation." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2023.](https://mlanthology.org/cvprw/2023/b2023cvprw-continual/) doi:10.1109/CVPRW59228.2023.00244

BibTeX

@inproceedings{b2023cvprw-continual,
  title     = {{Continual Domain Adaptation Through Pruning-Aided Domain-Specific Weight Modulation}},
  author    = {B, Prasanna and Sanyal, Sunandini and Babu, R. Venkatesh},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
  year      = {2023},
  pages     = {2457-2463},
  doi       = {10.1109/CVPRW59228.2023.00244},
  url       = {https://mlanthology.org/cvprw/2023/b2023cvprw-continual/}
}