A multi-view consistency framework with semi-supervised domain adaptation

Yuting Hong, Li Dong, Xiaojie Qiu, Hui Xiao, Baochen Yao, Siming Zheng, Chengbin Peng

Abstract: Highlights•Semi-supervised learning can reduce the cost for domain shift.•The multi-view approach adopts debiased pseudo-labels and pseudo-negative labels.•It reduces class ambiguity significantly.•The cross-domain affinity learning helps feature alignment across domains.•Our approach achieves good results on well-known domain adaptation datasets.