Adaptive margin for unsupervised domain adaptation without source data

Ziyun Cai, Yawen Huang, Tengfei Zhang, Changhui Hu, Xiao-Yuan Jing

Abstract: Highlights•A cutting-edge framework called AM-SFDA, which effectively addresses SFDA problem.•History of margins plays a crucial role in the stability of deep learning models.•The proposed method can achieve state-of-the-art performance under SFDA scenario.