TY - JOUR
T1 - Unsupervised Domain Adaptation via Domain-Adaptive Diffusion
AU - Peng, Duo
AU - Ke, Qiuhong
AU - Ambikapathi, Arul Murugan
AU - Yazici, Yasin
AU - Lei, Yinjie
AU - Liu, Jun
N1 - Publisher Copyright:
© 1992-2012 IEEE.
PY - 2024/7/15
Y1 - 2024/7/15
N2 - Unsupervised Domain Adaptation (UDA) is quite challenging due to the large distribution discrepancy between the source domain and the target domain. Inspired by diffusion models which have strong capability to gradually convert data distributions across a large gap, we consider to explore the diffusion technique to handle the challenging UDA task. However, using diffusion models to convert data distribution across different domains is a non-trivial problem as the standard diffusion models generally perform conversion from the Gaussian distribution instead of from a specific domain distribution. Besides, during the conversion, the semantics of the source-domain data needs to be preserved to classify correctly in the target domain. To tackle these problems, we propose a novel Domain-Adaptive Diffusion (DAD) module accompanied by a Mutual Learning Strategy (MLS), which can gradually convert data distribution from the source domain to the target domain while enabling the classification model to learn along the domain transition process. Consequently, our method successfully eases the challenge of UDA by decomposing the large domain gap into small ones and gradually enhancing the capacity of classification model to finally adapt to the target domain. Our method outperforms the current state-of-the-arts by a large margin on three widely used UDA datasets.
AB - Unsupervised Domain Adaptation (UDA) is quite challenging due to the large distribution discrepancy between the source domain and the target domain. Inspired by diffusion models which have strong capability to gradually convert data distributions across a large gap, we consider to explore the diffusion technique to handle the challenging UDA task. However, using diffusion models to convert data distribution across different domains is a non-trivial problem as the standard diffusion models generally perform conversion from the Gaussian distribution instead of from a specific domain distribution. Besides, during the conversion, the semantics of the source-domain data needs to be preserved to classify correctly in the target domain. To tackle these problems, we propose a novel Domain-Adaptive Diffusion (DAD) module accompanied by a Mutual Learning Strategy (MLS), which can gradually convert data distribution from the source domain to the target domain while enabling the classification model to learn along the domain transition process. Consequently, our method successfully eases the challenge of UDA by decomposing the large domain gap into small ones and gradually enhancing the capacity of classification model to finally adapt to the target domain. Our method outperforms the current state-of-the-arts by a large margin on three widely used UDA datasets.
KW - diffusion model
KW - image classification
KW - transfer learning
KW - Unsupervised domain adaptation
UR - http://www.scopus.com/inward/record.url?scp=85199057927&partnerID=8YFLogxK
U2 - 10.1109/TIP.2024.3424985
DO - 10.1109/TIP.2024.3424985
M3 - Article
C2 - 39008383
AN - SCOPUS:85199057927
SN - 1941-0042
VL - 33
SP - 4245
EP - 4260
JO - IEEE Transactions on Image Processing
JF - IEEE Transactions on Image Processing
ER -