Domain adaptation based on feature-level and class-level alignment

被引：0
|
作者：
Zhao X.-Q. [1 ,2 ,3 ]
Jiang H.-M. [1 ,2 ,3 ]
机构：
[1] College of Electrical and Information Engineering, Lanzhou University of Technology, Lanzhou
[2] Gansu Key Laboratory of Advanced Control for Industrial Processes, Lanzhou University of Technology, Lanzhou
[3] National Experimental Teaching Center of Electrical and Control Engineering, Lanzhou University of Technology, Lanzhou
来源：
Kongzhi yu Juece/Control and Decision | 2022年 / 37卷 / 05期
关键词：
Domain adaptation; Joint discriminant network; MID measurement function; Residual correction block;
D O I：
10.13195/j.kzyjc.2020.1241
中图分类号：
学科分类号：
摘要：
Aiming at the problems of existing domain adaptation algorithms based on adversarial learning that they cannot effectively learn transferable features and have poor generalization ability, a domain adaptation algorithm based on feature and category alignment(FCDA) is proposed in this paper. First of all, in view of the shortcomings of the maximum mean discrepancy(MMD) measurement criteria, a new improved maximizes the intra-domain density(MID) measurement function is obtained, which measures the distribution divergence between the source domain sample features with the same label, and the distribution divergence between the target domain sample features with the same label, so as to maximize the class density of similar samples in the domain, thereby the class error rate is reduced. Then, in order to learn the abstract and transferable features of the target sample at a deeper level, and reduce the difference between domains, a residual correction block is added after the feature extraction network to deepen the basic network, and the transferability of its features is improved. Finally, the acquired features are passed through the joint discriminant network, and the alignments at the class-level and the domain-level are achieved with the adversarial loss function. The proposed algorithm has an average accuracy of 88.6% for the dataset Office-31 and an average accuracy of 67.7% for the dataset Office-Home. Compared with other algorithms, the proposed algorithm has better generalization ability and higher classification performance. Copyright ©2022 Control and Decision.
引用
收藏
页码：1203 / 1210
页数：7
相关论文
共 25 条
  • [1] Long M S., Transfer learning: Problems and methods, (2014)
  • [2] Pan S J, Yang Q., A survey on transfer learning, IEEE Transactions on Knowledge and Data Engineering, 22, 10, pp. 1345-1359, (2010)
  • [3] Bousmalis K, Silberman N, Dohan D, Et al., Unsupervised pixel-level domain adaptation with generative adversarial networks, 2017 IEEE Conference on Computer Vision and Pattern Recognition, pp. 95-104, (2017)
  • [4] Lunga D, Yang H L, Reith A, Et al., Domain-adapted convolutional networks for satellite image classification: A large-scale interactive learning workflow, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 11, 3, pp. 962-977, (2018)
  • [5] Sohn K, Liu S F, Zhong G Y, Et al., Unsupervised domain adaptation for face recognition in unlabeled videos, Proceedings of the 2017 IEEE International Conference on Computer Vision, pp. 5917-5925, (2017)
  • [6] Chen Y, Li W, Sakaridis C, Et al., Domain adaptive faster R-CNN for object detection in the wild, Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pp. 3339-3348, (2018)
  • [7] Zou Y, Yu Z, Kumar B V K V, Et al., Unsupervised domain adaptation for semantic segmentation via class-balanced self-training, Proceedings of the European Conference on Computer Vision, pp. 297-313, (2018)
  • [8] Wang M, Deng W H., Deep visual domain adaptation: A survey, Neurocomputing, 312, pp. 135-153, (2018)
  • [9] Wang X S, Zhao J J, Cheng Y H., Unsupervised adaptive classifier based on extreme learning machines, Control and Decision, 35, 4, pp. 861-869, (2020)
  • [10] Zang S F, Cheng Y H, Wang X S., Domain adaptation learning based on maximum distribution weighted mean discrepancy, Control and Decision, 31, 11, pp. 2083-2089, (2016)
123