TY - GEN
T1 - Towards Better Robustness against Common Corruptions for Unsupervised Domain Adaptation
AU - Gao, Zhiqiang
AU - Huang, Kaizhu
AU - Zhang, Rui
AU - Liu, Dawei
AU - Ma, Jieming
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2024/1/15
Y1 - 2024/1/15
N2 - Recent studies have investigated how to achieve robustness for unsupervised domain adaptation (UDA). While most efforts focus on adversarial robustness, i.e. how the model performs against unseen malicious adversarial perturbations, robustness against benign common corruption (RaCC) surprisingly remains under-explored for UDA. Towards improving RaCC for UDA methods in an unsupervised manner, we propose a novel Distributionally and Discretely Adversarial Regularization (DDAR) framework in this paper. Formulated as a min-max optimization with a distribution distance, DDAR1 is theoretically well-founded to ensure generalization over unknown common corruptions. Meanwhile, we show that our regularization scheme effectively reduces a surrogate of RaCC, i.e., the perceptual distance between natural data and common corruption. To enable a abetter adversarial regularization, the design of the optimization pipeline relies on an image discretization scheme that can transform "out-of-distribution"adversarial data into "in-distribution"data augmentation. Through extensive experiments, in terms of RaCC, our method is superior to conventional unsupervised regularization mechanisms, widely improves the robustness of existing UDA methods, and achieves state-of-the-art performance.
AB - Recent studies have investigated how to achieve robustness for unsupervised domain adaptation (UDA). While most efforts focus on adversarial robustness, i.e. how the model performs against unseen malicious adversarial perturbations, robustness against benign common corruption (RaCC) surprisingly remains under-explored for UDA. Towards improving RaCC for UDA methods in an unsupervised manner, we propose a novel Distributionally and Discretely Adversarial Regularization (DDAR) framework in this paper. Formulated as a min-max optimization with a distribution distance, DDAR1 is theoretically well-founded to ensure generalization over unknown common corruptions. Meanwhile, we show that our regularization scheme effectively reduces a surrogate of RaCC, i.e., the perceptual distance between natural data and common corruption. To enable a abetter adversarial regularization, the design of the optimization pipeline relies on an image discretization scheme that can transform "out-of-distribution"adversarial data into "in-distribution"data augmentation. Through extensive experiments, in terms of RaCC, our method is superior to conventional unsupervised regularization mechanisms, widely improves the robustness of existing UDA methods, and achieves state-of-the-art performance.
UR - http://www.scopus.com/inward/record.url?scp=85183440907&partnerID=8YFLogxK
U2 - 10.1109/ICCV51070.2023.01731
DO - 10.1109/ICCV51070.2023.01731
M3 - Conference Proceeding
T3 - Proceedings of the IEEE International Conference on Computer Vision
SP - 18836
EP - 18847
BT - Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023
Y2 - 2 October 2023 through 6 October 2023
ER -