Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference

Liu, Xiaofeng; Hu, Bo; Jin, Linghao; Han, Xu; Xing, Fangxu; Ouyang, Jinsong; Lu, Jun; El Fakhri, Georges; Woo, Jonghye

doi:10.24963/ijcai.2021/122

Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference

Xiaofeng Liu, Bo Hu, Linghao Jin, Xu Han, Fangxu Xing, Jinsong Ouyang, Jun Lu, Georges El Fakhri, Jonghye Woo

Download BibTeX Cite

Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence. Main Track. Pages 881-887.

https://doi.org/10.24963/ijcai.2021/122

Keywords

Computer Vision: Recognition: Detection, Categorization, Indexing, Matching, Retrieval, Semantic Interpretation
Computer Vision: Statistical Methods and Machine Learning
Machine Learning: Classification

Abstract

In this work, we propose a domain generalization (DG) approach to learn on several labeled source domains and transfer knowledge to a target domain that is inaccessible in training. Considering the inherent conditional and label shifts, we would expect the alignment of p(x|y) and p(y). However, the widely used domain invariant feature learning (IFL) methods relies on aligning the marginal concept shift w.r.t. p(x), which rests on an unrealistic assumption that p(y) is invariant across domains. We thereby propose a novel variational Bayesian inference framework to enforce the conditional distribution alignment w.r.t. p(x|y) via the prior distribution matching in a latent space, which also takes the marginal label shift w.r.t. p(y) into consideration with the posterior alignment. Extensive experiments on various benchmarks demonstrate that our framework is robust to the label shift and the cross-domain accuracy is significantly improved, thereby achieving superior performance over the conventional IFL counterparts.

Xiaofeng Liu, Bo Hu, Linghao Jin, Xu Han, Fangxu Xing, Jinsong Ouyang, Jun Lu, Georges El Fakhri, and Jonghye Woo. 2021. Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21. IJCAI, 881–887. https://doi.org/10.24963/ijcai.2021/122

@inproceedings{ijcai2021p122,
  title     = {Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference},
  author    = {Liu, Xiaofeng and Hu, Bo and Jin, Linghao and Han, Xu and Xing, Fangxu and Ouyang, Jinsong and Lu, Jun and El Fakhri, Georges and Woo, Jonghye},
  booktitle = {Proceedings of the Thirtieth International Joint Conference on
               Artificial Intelligence, {IJCAI-21}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Prof. Zhi-Hua Zhou},
  pages     = {881--887},
  year      = {2021},
  month     = {8},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2021/122},
  url       = {https://doi.org/10.24963/ijcai.2021/122},
}

TY  - CPAPER
TI  - Domain Generalization under Conditional and Label Shifts via Variational Bayesian Inference
AU  - Liu, Xiaofeng
AU  - Hu, Bo
AU  - Jin, Linghao
AU  - Han, Xu
AU  - Xing, Fangxu
AU  - Ouyang, Jinsong
AU  - Lu, Jun
AU  - El Fakhri, Georges
AU  - Woo, Jonghye
T2  - Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21
PB  - International Joint Conferences on Artificial Intelligence Organization
PY  - 2021
SP  - 881
EP  - 887
DO  - 10.24963/ijcai.2021/122
UR  - https://doi.org/10.24963/ijcai.2021/122
ER  -