Nearest neighborhood-based deep clustering for source data-absent unsupervised domain adaptation

作     者：Tang, Song Yang, Yan Ma, Zhiyuan Hendrich, Norman Zeng, Fanyu Ge, Shuzhi Sam Zhang, Changshui Zhang, Jianwei 

作者机构：The Institute of Machine Intelligence University of Shanghai for Science and Technology Shanghai China The State Key Laboratory of Electronic Thin Films and Integrated Devices University of Electronic Science and Technology of China Chengdu China Group Department of Informatics Universität Hamburg Hamburg Germany The Institute of Machine Intelligence University of Shanghai for Science and Technology Shanghai China The State Key Lab. for Novel Software Technology Nanjing University Nanjing China The Engineering Research Center of Wideband Wireless Communication Technology Ministry of Education Nanjing University of Posts and Telecommunications Nanjing China The Department of Electrical and Computer Engineering National University of Singapore Singapore Singapore The Department of Automation Tsinghua University Beijing China 

出 版 物：《arXiv》 (arXiv)

年 卷 期：2021年

核心收录：

主　　题：Semantics 

摘      要：In the classic setting of unsupervised domain adaptation (UDA), the labeled source data are available in the training phase. However, in many real-world scenarios, owing to some reasons such as privacy protection and information security, the source data is inaccessible, and only a model trained on the source domain is available. This paper proposes a novel deep clustering method for this challenging task. Aiming at the dynamical clustering at feature-level, we introduce extra constraints hidden in the geometric structure between data to assist the process. Concretely, we propose a geometry-based constraint, named semantic consistency on the nearest neighborhood (SCNNH), and use it to encourage robust clustering. To reach this goal, we construct the nearest neighborhood for every target data and take it as the fundamental clustering unit by building our objective on the geometry. Also, we develop a more SCNNH-compliant structure with an additional semantic credibility constraint, named semantic hyper-nearest neighborhood (SHNNH). After that, we extend our method to this new geometry. Extensive experiments on three challenging UDA datasets indicate that our method achieves state-of-the-art results. The proposed method has significant improvement on all datasets (as we adopt SHNNH, the average accuracy increases by over 3.0% on the large-scaled dataset). Code is available at https://***/tntek/N2DCX. Copyright © 2021, The Authors. All rights reserved.