Cross-Domain Gradient Discrepancy Minimization for Unsupervised Domain Adaptation

被引:109
|
作者
Du, Zhekai [1 ]
Li, Jingjing [1 ]
Su, Hongzu [1 ]
Zhu, Lei [2 ]
Lu, Ke [1 ]
机构
[1] Univ Elect Sci & Technol China, Chengdu, Sichuan, Peoples R China
[2] Shandong Normal Univ, Jinan, Shandong, Peoples R China
基金
中国国家自然科学基金;
关键词
D O I
10.1109/CVPR46437.2021.00393
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Unsupervised Domain Adaptation (UDA) aims to generalize the knowledge learned from a well-labeled source domain to an unlabled target domain. Recently, adversarial domain adaptation with two distinct classifiers (bi-classifier) has been introduced into UDA which is effective to align distributions between different domains. Previous bi-classifier adversarial learning methods only focus on the similarity between the outputs of two distinct classifiers. However, the similarity of the outputs cannot guarantee the accuracy of target samples, i.e., traget samples may match to wrong categories even if the discrepancy between two classifiers is small. To challenge this issue, in this paper, we propose a cross-domain gradient discrepancy minimization (CGDM) method which explicitly minimizes the discrepancy of gradients generated by source samples and target samples. Specifically, the gradient gives a cue for the semantic information of target samples so it can be used as a good supervision to improve the accuracy of target samples. In order to compute the gradient signal of target smaples, we further obtain target pseudo labels through a clustering-based self-supervised learning. Extensive experiments on three widely used UDA datasets show that our method surpasses many previous state-of-the-arts.
引用
收藏
页码:3936 / 3945
页数:10
相关论文
共 50 条
  • [1] Cross-Domain Error Minimization for Unsupervised Domain Adaptation
    Du, Yuntao
    Chen, Yinghao
    Cui, Fengli
    Zhang, Xiaowen
    Wang, Chongjun
    [J]. DATABASE SYSTEMS FOR ADVANCED APPLICATIONS (DASFAA 2021), PT II, 2021, 12682 : 429 - 448
  • [2] Unsupervised domain adaptation through adversarial enhancement and gradient discrepancy minimization
    Tian, Qing
    Yang, Hong
    Lu, Zhanghu
    Liu, Mengyu
    [J]. COMPUTERS & ELECTRICAL ENGINEERING, 2023, 105
  • [3] Unsupervised Domain Adaptation with Imbalanced Cross-Domain Data
    Hsu, Tzu-Ming Harry
    Chen, Wei-Yu
    Hou, Cheng-An
    Tsai, Yao-Hung Hubert
    Yeh, Yi-Ren
    Wang, Yu-Chiang Frank
    [J]. 2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, : 4121 - 4129
  • [4] Cross-domain feature enhancement for unsupervised domain adaptation
    Long Sifan
    Wang Shengsheng
    Zhao Xin
    Fu Zihao
    Wang Bilin
    [J]. Applied Intelligence, 2022, 52 : 17326 - 17340
  • [5] Cross-Domain Contrastive Learning for Unsupervised Domain Adaptation
    Wang, Rui
    Wu, Zuxuan
    Weng, Zejia
    Chen, Jingjing
    Qi, Guo-Jun
    Jiang, Yu-Gang
    [J]. IEEE TRANSACTIONS ON MULTIMEDIA, 2023, 25 : 1665 - 1673
  • [6] Cross-domain feature enhancement for unsupervised domain adaptation
    Sifan, Long
    Shengsheng, Wang
    Xin, Zhao
    Zihao, Fu
    Bilin, Wang
    [J]. APPLIED INTELLIGENCE, 2022, 52 (15) : 17326 - 17340
  • [7] Adversarial style discrepancy minimization for unsupervised domain adaptation
    Luo, Xin
    Chen, Wei
    Liang, Zhengfa
    Li, Chen
    Tan, Yusong
    [J]. NEURAL NETWORKS, 2023, 157 : 216 - 225
  • [8] FeatureTransfer: Unsupervised Domain Adaptation for Cross-Domain Deepfake Detection
    Chen, Baoying
    Tan, Shunquan
    [J]. SECURITY AND COMMUNICATION NETWORKS, 2021, 2021
  • [9] Unsupervised Domain Adaptation for Cross-domain Histopathology Image Classification
    Li, Xiangning
    Pan, Chen
    He, Lingmin
    Li, Xinyu
    [J]. Multimedia Tools and Applications, 2024, 83 (08) : 23311 - 23331
  • [10] Cross-Domain Graph Convolutions for Adversarial Unsupervised Domain Adaptation
    Zhu, Ronghang
    Jiang, Xiaodong
    Lu, Jiasen
    Li, Sheng
    [J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2023, 34 (08) : 3847 - 3858