Reusing the Task-specific Classifier as a Discriminator: Discriminator-free Adversarial Domain Adaptation

被引:100
作者
Chen, Lin [1 ]
Chen, Huaian [1 ]
Wei, Zhixiang [1 ]
Jin, Xin [1 ]
Tan, Xiao [1 ]
Jin, Yi [1 ]
Chen, Enhong [1 ]
机构
[1] Univ Sci & Technol China, Hefei, Peoples R China
来源
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) | 2022年
基金
中国国家自然科学基金;
关键词
D O I
10.1109/CVPR52688.2022.00704
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Adversarial learning has achieved remarkable performances for unsupervised domain adaptation (UDA). Existing adversarial UDA methods typically adopt an additional discriminator to play the min-max game with a feature extractor. However, most of these methods failed to effectively leverage the predicted discriminative information, and thus cause mode collapse for generator. In this work, we address this problem from a different perspective and design a simple yet effective adversarial paradigm in the form of a discriminator-free adversarial learning network (DALN), wherein the category classifier is reused as a discriminator, which achieves explicit domain alignment and category distinguishment through a unified objective, enabling the DALN to leverage the predicted discriminative information for sufficient feature alignment. Basically, we introduce a Nuclear-norm Wasserstein discrepancy (NWD) that has definite guidance meaning for performing discrimination. Such NWD can be coupled with the classifier to serve as a discriminator satisfying the K-Lipschitz constraint without the requirements of additional weight clipping or gradient penalty strategy. Without bells and whistles, DALN compares favorably against the existing state-of-the-art (SOTA) methods on a variety of public datasets. Moreover, as a plug-and-play technique, NWD can be directly used as a generic regularizer to benefit existing UDA algorithms. Code is available at https://github.com/xiaoachen98/DALN.
引用
收藏
页码:7171 / 7180
页数:10
相关论文
共 51 条
  • [1] [Anonymous], 2004, NIPS
  • [2] Arjovsky M, 2017, PR MACH LEARN RES, V70
  • [3] Ben-David S., 2007, P ADV NEUR INF PROC, P137, DOI 10.7551/mitpress/7503.003.0022
  • [4] Carion Nicolas, 2020, Computer Vision - ECCV 2020. 16th European Conference. Proceedings. Lecture Notes in Computer Science (LNCS 12346), P213, DOI 10.1007/978-3-030-58452-8_13
  • [5] DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
    Chen, Liang-Chieh
    Papandreou, George
    Kokkinos, Iasonas
    Murphy, Kevin
    Yuille, Alan L.
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2018, 40 (04) : 834 - 848
  • [6] Chen XY, 2019, PR MACH LEARN RES, V97
  • [7] Cui SH, 2021, Arxiv, DOI arXiv:2107.06154
  • [8] Towards Discriminability and Diversity: Batch Nuclear-norm Maximization under Label Insufficient Situations
    Cui, Shuhao
    Wang, Shuhui
    Zhuo, Junbao
    Li, Liang
    Huang, Qingming
    Tian, Qi
    [J]. 2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2020, : 3940 - 3949
  • [9] Gradually Vanishing Bridge for Adversarial Domain Adaptation
    Cui, Shuhao
    Wang, Shuhui
    Zhuo, Junbao
    Su, Chi
    Huang, Qingming
    Tian, Qi
    [J]. 2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2020), 2020, : 12452 - 12461
  • [10] Cross-Domain Gradient Discrepancy Minimization for Unsupervised Domain Adaptation
    Du, Zhekai
    Li, Jingjing
    Su, Hongzu
    Zhu, Lei
    Lu, Ke
    [J]. 2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, : 3936 - 3945