Metric learning for domain adversarial network  

作　　者：Haifeng HU Yan YANG Yueming YIN Jiansheng WU 

机构地区：[1]College of Telecommunications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210023,China [2]College of Geography and Biological Information,Nanjing University of Posts and Telecommunications,Nanjing 210023,China

出　　处：《Frontiers of Computer Science》2022年第5期229-231,共3页中国计算机科学前沿（英文版）

基　　金：This work was supported in part by the National Natural Science Foundation of China(Grant Nos.62071242,61571233,61901229,and 61872198);the Graduate Research and Innovation Projects of Jiangsu Province(KYCX20_0738).

摘　　要：1 Introduction The existing domain adaptation methods[1,2]always aim to perform domain alignment between the source and target domain to alleviate the problem of domain shift[3].The target domain samples are likely to scatter on the classification boundary due to a lack of label information.Therefore,how to identify these overlapping classes in the target domain,called easily-confused classes,becomes the key to the improvement of classification performance.

关 键 词：FUSED ALIGNMENT SCATTER 

分 类 号：O17[理学—数学]