arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.02191cs.LGcs.AIcs.CRmath.STstat.MLstat.TH
收藏

针对差分隐私的攻击感知噪声校准

Attack-Aware Noise Calibration for Differential Privacy

Lausanne University Hospital (CHUV)(洛桑大学医院(CHUV));Harvard University(哈佛大学);Technical University Munich(慕尼黑工业大学);EPFL(洛桑联邦理工学院)

Bogdan Kulynych, Juan Felipe Gomez, Georgios Kaissis, Flavio du Pin Calmon, Carmela Troncoso

AI总结:

针对差分隐私,提出直接按攻击风险校准噪声规模的方法,替代先选ε再转换,显著降低噪声、提升效用,实验证明在相同风险下提高模型准确率。

摘要:

差分隐私(DP)是一种广泛使用的方法,用于在敏感数据上训练机器学习模型时降低隐私风险。DP机制在训练过程中添加噪声,以限制信息泄露的风险。所添加噪声的规模至关重要,因为它决定了隐私与效用之间的权衡。标准做法是选择噪声规模以满足给定的隐私预算ε。该隐私预算又根据操作层面的攻击风险来解释,例如旨在恢复训练数据记录信息的推理攻击的准确率、敏感性和特异性。我们表明,首先将噪声规模校准到隐私预算ε,然后将ε转换为攻击风险,会导致过于保守的风险评估和不必要的低效用。相反,我们提出了直接将噪声规模校准到期望攻击风险水平的方法,绕过了选择ε的步骤。对于给定的攻击风险概念,我们的方法显著降低了噪声规模,从而在相同隐私水平下提高了效用。我们通过实验证明,在训练隐私保护的机器学习模型时,将噪声校准到攻击敏感性/特异性(而非ε)能大幅提高相同风险水平下的模型准确率。我们的工作提供了一种原则性且实用的方法,在不损害隐私的情况下提高隐私保护机器学习的效用。代码可在https://github.com/Felipe-Gomez/riskcal获取。

英文摘要:

Differential privacy (DP) is a widely used approach for mitigating privacy risks when training machine learning models on sensitive data. DP mechanisms add noise during training to limit the risk of information leakage. The scale of the added noise is critical, as it determines the trade-off between privacy and utility. The standard practice is to select the noise scale to satisfy a given privacy budget $\varepsilon$. This privacy budget is in turn interpreted in terms of operational attack risks, such as accuracy, sensitivity, and specificity of inference attacks aimed to recover information about the training data records. We show that first calibrating the noise scale to a privacy budget $\varepsilon$, and then translating ε to attack risk leads to overly conservative risk assessments and unnecessarily low utility. Instead, we propose methods to directly calibrate the noise scale to a desired attack risk level, bypassing the step of choosing $\varepsilon$. For a given notion of attack risk, our approach significantly decreases noise scale, leading to increased utility at the same level of privacy. We empirically demonstrate that calibrating noise to attack sensitivity/specificity, rather than $\varepsilon$, when training privacy-preserving ML models substantially improves model accuracy for the same risk level. Our work provides a principled and practical way to improve the utility of privacy-preserving ML without compromising on privacy. The code is available at https://github.com/Felipe-Gomez/riskcal

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑