TY - GEN
T1 - Joint adversarial domain adaptation
AU - Li, Shuang
AU - Liu, Chi Harold
AU - Xie, Binhui
AU - Su, Limin
AU - Ding, Zhengming
AU - Huang, Gao
N1 - Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019/10/15
Y1 - 2019/10/15
N2 - Domain adaptation aims to transfer the enriched label knowledge from large amounts of source data to unlabeled target data. It has raised significant interest in multimedia analysis. Existing researches mainly focus on learning domain-wise transferable representations via statistical moment matching or adversarial adaptation techniques, while ignoring the class-wise mismatch across domains, resulting in inaccurate distribution alignment. To address this issue, we propose a Joint Adversarial Domain Adaptation (JADA) approach to simultaneously align domain-wise and class-wise distributions across source and target in a unified adversarial learning process. Specifically, JADA attempts to solve two complementary minimax problems jointly. The feature generator aims to not only fool the well-trained domain discriminator to learn domain-invariant features, but also minimize the disagreement between two distinct task-specific classifiers' predictions to synthesize target features near the support of source class-wisely. As a result, the learned transferable features will be equipped with more discriminative structures, and effectively avoid mode collapse. Additionally, JADA enables an efficient end-to-end training manner via a simple back-propagation scheme. Extensive experiments on several real-world cross-domain benchmarks, including VisDA-2017, ImageCLEF, Office-31 and digits, verify that JADA can gain remarkable improvements over other state-of-the-art deep domain adaptation approaches.
AB - Domain adaptation aims to transfer the enriched label knowledge from large amounts of source data to unlabeled target data. It has raised significant interest in multimedia analysis. Existing researches mainly focus on learning domain-wise transferable representations via statistical moment matching or adversarial adaptation techniques, while ignoring the class-wise mismatch across domains, resulting in inaccurate distribution alignment. To address this issue, we propose a Joint Adversarial Domain Adaptation (JADA) approach to simultaneously align domain-wise and class-wise distributions across source and target in a unified adversarial learning process. Specifically, JADA attempts to solve two complementary minimax problems jointly. The feature generator aims to not only fool the well-trained domain discriminator to learn domain-invariant features, but also minimize the disagreement between two distinct task-specific classifiers' predictions to synthesize target features near the support of source class-wisely. As a result, the learned transferable features will be equipped with more discriminative structures, and effectively avoid mode collapse. Additionally, JADA enables an efficient end-to-end training manner via a simple back-propagation scheme. Extensive experiments on several real-world cross-domain benchmarks, including VisDA-2017, ImageCLEF, Office-31 and digits, verify that JADA can gain remarkable improvements over other state-of-the-art deep domain adaptation approaches.
KW - Adversarial Learning
KW - Back-propagation
KW - Domain Adaptation
KW - Gradient Reversal Layer
KW - Representation Learning
UR - http://www.scopus.com/inward/record.url?scp=85074844617&partnerID=8YFLogxK
U2 - 10.1145/3343031.3351070
DO - 10.1145/3343031.3351070
M3 - Conference contribution
AN - SCOPUS:85074844617
T3 - MM 2019 - Proceedings of the 27th ACM International Conference on Multimedia
SP - 729
EP - 737
BT - MM 2019 - Proceedings of the 27th ACM International Conference on Multimedia
PB - Association for Computing Machinery, Inc
T2 - 27th ACM International Conference on Multimedia, MM 2019
Y2 - 21 October 2019 through 25 October 2019
ER -