Domain Adaptation with Adversarial Training on Penultimate Activations

被引:0
|
作者
Sun, Tao [1 ]
Lu, Cheng [2 ]
Ling, Haibin [1 ]
机构
[1] SUNY Stony Brook, Stony Brook, NY 11794 USA
[2] XPeng Motors, Guangzhou, Peoples R China
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Enhancing model prediction confidence on target data is an important objective in Unsupervised Domain Adaptation (UDA). In this paper, we explore adversarial training on penultimate activations, i.e., input features of the final linear classification layer. We show that this strategy is more efficient and better correlated with the objective of boosting prediction confidence than adversarial training on input images or intermediate features, as used in previous works. Furthermore, with activation normalization commonly used in domain adaptation to reduce domain gap, we derive two variants and systematically analyze the effects of normalization on our adversarial training. This is illustrated both in theory and through empirical analysis on real adaptation tasks. Extensive experiments are conducted on popular UDA benchmarks under both standard setting and source-data free setting. The results validate that our method achieves the best scores against previous arts. Code is available at https://github.com/tsun/APA.
引用
收藏
页码:9935 / 9943
页数:9
相关论文
共 50 条
  • [21] Self-training Guided Adversarial Domain Adaptation For Thermal Imagery
    Akkaya, Ibrahim Batuhan
    Altinel, Fazil
    Halici, Ugur
    2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS, CVPRW 2021, 2021, : 4317 - 4326
  • [22] Pairwise Adversarial Training for Unsupervised Class-imbalanced Domain Adaptation
    Shi, Weili
    Zhu, Ronghang
    Li, Sheng
    PROCEEDINGS OF THE 28TH ACM SIGKDD CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, KDD 2022, 2022, : 1598 - 1606
  • [23] Low Curvature Activations Reduce Overfitting in Adversarial Training
    Singla, Vasu
    Singla, Sahil
    Feizi, Soheil
    Jacobs, David
    2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 16403 - 16413
  • [24] Active Adversarial Domain Adaptation
    Su, Jong-Chyi
    Sai, Yi-Hsuan
    Sohn, Kihyuk
    Liu, Buyu
    Maji, Subhransu
    Chandraker, Manmohan
    2020 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV), 2020, : 728 - 737
  • [25] Adversarial Discriminative Domain Adaptation
    Tzeng, Eric
    Hoffman, Judy
    Saenko, Kate
    Darrell, Trevor
    30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017), 2017, : 2962 - 2971
  • [26] Partial Adversarial Domain Adaptation
    Cao, Zhangjie
    Ma, Lijia
    Long, Mingsheng
    Wang, Jianmin
    COMPUTER VISION - ECCV 2018, PT VIII, 2018, 11212 : 139 - 155
  • [27] A Survey on Adversarial Domain Adaptation
    Zonoozi, Mahta HassanPour
    Seydi, Vahid
    NEURAL PROCESSING LETTERS, 2023, 55 (03) : 2429 - 2469
  • [28] A Survey on Adversarial Domain Adaptation
    Mahta HassanPour Zonoozi
    Vahid Seydi
    Neural Processing Letters, 2023, 55 : 2429 - 2469
  • [29] Conditional Adversarial Domain Adaptation
    Long, Mingsheng
    Cao, Zhangjie
    Wang, Jianmin
    Jordan, Michael I.
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 31 (NIPS 2018), 2018, 31
  • [30] Discriminative Adversarial Domain Adaptation
    Tang, Hui
    Jia, Kui
    THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 5940 - 5947