FixBi: bridging domain spaces for unsupervised domain adaptation

Jaemin Na, Heechul Jung, Hyung Jin Chang, Wonjun Hwang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

56 Downloads (Pure)

Abstract

Unsupervised domain adaptation (UDA) methods for learning domain invariant representations have achieved remarkable progress. However, most of the studies were based on direct adaptation from the source domain to the target domain and have suffered from large domain discrepancies. In this paper, we propose a UDA method that effectively handles such large domain discrepancies. We introduce a fixed ratio-based mixup to augment multiple intermediate domains between the source and target domain. From the augmented-domains, we train the source-dominant model and the target-dominant model that have complementary characteristics. Using our confidence based learning methodologies, e.g., bidirectional matching with high-confidence predictions and self-penalization using low-confidence predictions, the models can learn from each other or from its own results. Through our proposed methods, the models gradually transfer domain knowledge from the source to the target domain. Extensive experiments demonstrate the superiority of our proposed method on three public benchmarks: Office-31, Office-Home, and VisDA-2017.
Original languageEnglish
Title of host publication2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
PublisherIEEE
Number of pages10
ISBN (Electronic)978-1-6654-4509-2
ISBN (Print)978-1-6654-4510-8
DOIs
Publication statusPublished - 2 Nov 2021
Event2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition - Nashville, United States
Duration: 21 Jun 202124 Jun 2021

Publication series

NameProceedings. IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
PublisherIEEE
ISSN (Electronic)2575-7075

Conference

Conference2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition
Abbreviated titleCVPR 2021
Country/TerritoryUnited States
CityNashville
Period21/06/2124/06/21

Keywords

  • computer vision
  • adaptation models
  • benchmark testing
  • predictive models
  • pattern recognition
  • standards

Fingerprint

Dive into the research topics of 'FixBi: bridging domain spaces for unsupervised domain adaptation'. Together they form a unique fingerprint.

Cite this