《電子技術(shù)應(yīng)用》
您所在的位置:首頁(yè) > 其他 > 設(shè)計(jì)應(yīng)用 > 依賴差分隱私:關(guān)聯(lián)數(shù)據(jù)集下的高斯機(jī)制
依賴差分隱私:關(guān)聯(lián)數(shù)據(jù)集下的高斯機(jī)制
網(wǎng)絡(luò)安全與數(shù)據(jù)治理
歐陽(yáng)恒,陳洪超
貴州輕工職業(yè)技術(shù)學(xué)院信息工程系
摘要: 差分隱私(Differential Privacy)是一種數(shù)據(jù)擾動(dòng)框架,它保證查詢結(jié)果在概率上不可區(qū)分。研究表明差分隱私應(yīng)用于關(guān)聯(lián)數(shù)據(jù)集時(shí),將帶來(lái)隱私泄露的風(fēng)險(xiǎn)。根據(jù)依賴差分隱私(Dependent Differential Privacy),量化了依賴差分隱私敏感度的度量;隨后,提出了依賴差分隱私-高斯機(jī)制算法(Gaussian Mechanism Algorithm Dependent Differential Privacy),實(shí)現(xiàn)數(shù)據(jù)擾動(dòng),同時(shí)證明了該機(jī)制滿足隱私保證的基本定理;通過(guò)使用真實(shí)數(shù)據(jù)集的實(shí)驗(yàn)表明,GMA DDP在管理依賴數(shù)據(jù)的隱私-效用權(quán)衡方面具有較高的可用性。
中圖分類號(hào):TP309.2文獻(xiàn)標(biāo)識(shí)碼:ADOI:10.19358/j.issn.2097-1788.2024.03.002
引用格式:歐陽(yáng)恒,陳洪超.依賴差分隱私:關(guān)聯(lián)數(shù)據(jù)集下的高斯機(jī)制[J].網(wǎng)絡(luò)安全與數(shù)據(jù)治理,2024,43(3):9-13.
Dependent differential privacy: Gaussian mechanism for correlated datasets
OuYang Heng, Chen Hongchao
Department of Information Engineering, Guizhou Light Industry Technical College
Abstract: Differential Privacy is a data perturbation framework, which ensures that the query results are not distinguishable in probability. Research shows that when differential privacy is applied to associated data sets, it will bring the risk of privacy disclosure. Based on the dependent differential privacy, this paper quantifies the sensitivity of the dependent differential privacy; Then, a Gaussian Mechanism Algorithm Dependent Differential Privacy is proposed to realize data disturbance, and the basic theorem that the mechanism meets the privacy guarantee is proved; Experiments using real data sets show that GMA DDP has high availability in managing privacy utility tradeoffs that depend on data.
Key words : differential privacy; dependent differential Privacy; Gaussian mechanism; correlated dataset

引言

數(shù)據(jù)成為信息時(shí)代最重要的生產(chǎn)要素,將帶來(lái)巨大的經(jīng)濟(jì)效益。然而,隨著數(shù)據(jù)分析技術(shù)與機(jī)器學(xué)習(xí)的發(fā)展,直接發(fā)布不經(jīng)過(guò)隱私保護(hù)處理的數(shù)據(jù),可能會(huì)導(dǎo)致隱私的泄露。Dwork等人[1]提出了差分隱私,作為一種擁有嚴(yán)格的數(shù)學(xué)定義和邏輯證明的隱私保護(hù)方法,能夠?yàn)閿?shù)據(jù)的發(fā)布提供強(qiáng)有力的隱私保護(hù)。高斯機(jī)制最初也由Dwork等人[1]提出,添加噪聲量σ2006最少應(yīng)滿足:σ2006≥Δ2log2/δ/ε,其中Δ是查詢的敏感度。然而,由于該方法噪聲量較大,沒(méi)有被廣泛應(yīng)用。隨后,Dwork等人[2]提出了一種優(yōu)化后的噪聲量計(jì)算方法σCGM≥Δ2log125/δ/ε,現(xiàn)已被廣泛采用。然而,2018年Balle等人[3]重新審視了高斯機(jī)制,提出了分析高斯機(jī)制(AGM)在噪聲量σAGM達(dá)到了最優(yōu)的效用,但由于其沒(méi)有封閉的表達(dá)式,需要使用二分法迭代計(jì)算,時(shí)間復(fù)雜度較高Θ(log2n)。


本文詳細(xì)內(nèi)容請(qǐng)下載:

http://ihrv.cn/resource/share/2000005928


作者信息:

歐陽(yáng)恒,陳洪超

貴州輕工職業(yè)技術(shù)學(xué)院信息工程系, 貴州貴陽(yáng)550025


雜志訂閱.jpg

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