Divide and augment: Supervised domain adaptation via sample-wise feature fusion

Zhuangzhuang Chen, Bin Pu, Lei Zhao, Jie He, Pengchen Liang

Abstract: Highlights•Addressing supervised domain adaptation via a sample-wise feature fusion rule.•Learning the decomposed features with the theoretical guarantee.•Transforming source samples into the target domain for target domain augmentation.•Our method performs favorably against existing supervised domain adaptation methods.