面向分类任务的隐私保护协作学习技术
网络安全与数据治理 2023年第5期
黎兰兰,张信明
(中国科学技术大学计算机学院,安徽合肥230026)
摘要: 随着相关法律法规的发布和人们隐私意识的觉醒,隐私保护要求不断提高。针对分类任务场景,提出了一种隐私性与效用性并重的面向分类任务的隐私保护协作技术(PCTC)。在隐私安全方面,将同态加密和差分隐私相结合,并设计了一种安全聚合机制,实现更加健壮的隐私保护;在效用性方面,引入信息熵的计算因子对标签可信度进行度量,降低标注噪声对模型性能的影响。最后对所提出的方案进行了安全性分析,并在公开数据集上进行了实验验证。结果表明PCTC在保证数据隐私安全的同时大幅度提升了模型性能,具有较为广阔的应用前景。
中圖分類號:TP393
文獻(xiàn)標(biāo)識碼:A
DOI:10.19358/j.issn.2097-1788.2023.05.007
引用格式:黎蘭蘭,張信明.面向分類任務(wù)的隱私保護(hù)協(xié)作學(xué)習(xí)技術(shù)[J].網(wǎng)絡(luò)安全與數(shù)據(jù)治理,2023,42(5):36-43.
文獻(xiàn)標(biāo)識碼:A
DOI:10.19358/j.issn.2097-1788.2023.05.007
引用格式:黎蘭蘭,張信明.面向分類任務(wù)的隱私保護(hù)協(xié)作學(xué)習(xí)技術(shù)[J].網(wǎng)絡(luò)安全與數(shù)據(jù)治理,2023,42(5):36-43.
Privacy-preserving collaborative learning technology for classification
Li Lanlan, Zhang Xinming
(School of Computer Science and Technology, University of Science and Technology of China, Hefei 230026, China)
Abstract: With the release of relevant laws and regulations and the awakening of people’s privacy awareness, the requirements for privacy protection are constantly increasing. Aiming at the scenario of classification, this paper proposes a Privacypreserving Collaborative Learning Technology for Classification (PCTC) that emphasizes both privacy and utility. In terms of privacy, homomorphic encryption and differential privacy are combined and a secure aggregation mechanism is designed to achieve more robust privacy protection. In terms of utility, the calculation factor of information entropy is introduced to measure the credibility of labels, which can reduce the impact of labeling noise on model performance. Finally, the security analysis of the proposed scheme is carried out, and the experiments are implemented on public datasets. The results show that PCTC significantly improves model performance while ensuring privacy and security of the data, and has broad application prospects.
Key words : privacy preservation; data labeling; classification task; homomorphic encryption; differential privacy
0 引言
近年來,隨著數(shù)據(jù)產(chǎn)生速度和計(jì)算機(jī)算力的持續(xù)提升,機(jī)器學(xué)習(xí)在目標(biāo)識別、語音識別、自然語言處理和對象檢測等許多領(lǐng)域都取得了巨大突破。新興的機(jī)器學(xué)習(xí)尤其是深度學(xué)習(xí)更是為產(chǎn)業(yè)的升級和變革提供了推動(dòng)力量,其中包括智慧農(nóng)業(yè)、智慧醫(yī)療等行業(yè)。良好的機(jī)器學(xué)習(xí)框架特別是有監(jiān)督的人工神經(jīng)網(wǎng)絡(luò)往往需要大量的標(biāo)注數(shù)據(jù),然而現(xiàn)實(shí)中任何單一實(shí)體都不可能總是擁有全部標(biāo)注數(shù)據(jù),多方協(xié)作學(xué)習(xí)是解決這一問題的有效方案。
本文詳細(xì)內(nèi)容請下載:http://ihrv.cn/resource/share/2000005332
作者信息:
黎蘭蘭,張信明
(中國科學(xué)技術(shù)大學(xué)計(jì)算機(jī)學(xué)院,安徽合肥230026)

此內(nèi)容為AET網(wǎng)站原創(chuàng),未經(jīng)授權(quán)禁止轉(zhuǎn)載。
