详细信息
Unsupervised Domain Adaptation with Differentially Private Gradient Projection ( SCI-EXPANDED收录 EI收录)
文献类型:期刊文献
英文题名:Unsupervised Domain Adaptation with Differentially Private Gradient Projection
作者:Zheng, Maobo[1];Zhang, Xiaojian[2];Ma, Xuebin[1]
第一作者:Zheng, Maobo
通讯作者:Ma, XB[1]
机构:[1]Inner Mongolia Univ, Inner Mongolia Key Lab Wireless Networking & Mobil, Hohhot, Peoples R China;[2]Henan Univ Econ & Law, Sch Comp & Informat Engn, Zhengzhou, Peoples R China
第一机构:Inner Mongolia Univ, Inner Mongolia Key Lab Wireless Networking & Mobil, Hohhot, Peoples R China
通讯机构:[1]corresponding author), Inner Mongolia Univ, Inner Mongolia Key Lab Wireless Networking & Mobil, Hohhot, Peoples R China.
年份:2023
卷号:2023
外文期刊名:INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
收录:;EI(收录号:20234615066544);Scopus(收录号:2-s2.0-85162704658);WOS:【SCI-EXPANDED(收录号:WOS:000965195400001)】;
基金:AcknowledgmentsThis study was supported by the Science and Technology Program of Inner Mongolia Autonomous Region (Grant no. 2019GG116) and the Engineering Research Center of Ecological Big Data, Ministry of Education.
语种:英文
摘要:Domain adaptation is a viable solution for deep learning with small data. However, domain adaptation models trained on data with sensitive information may be a violation of personal privacy. In this article, we proposed a solution for unsupervised domain adaptation, called DP-CUDA, which is based on differentially private gradient projection and contradistinguisher. Compared with the traditional domain adaptation process, DP-CUDA involves searching for domain-invariant features between the source domain and target domain first and then transferring knowledge. Specifically, the model is trained in the source domain by supervised learning from labeled data. During the training of the target model, feature learning is used to solve the classification task in an end-to-end manner using unlabeled data directly, and the differentially private noise is injected into the gradient. We conducted extensive experiments on a variety of benchmark datasets, including MNIST, USPS, SVHN, VisDA-2017, Office-31, and Amazon Review, to demonstrate our proposed method's utility and privacy-preserving properties.
参考文献:
正在载入数据...