Reconciling privacy an...
Artifcial intelligence (AI) models are vulnerable to information leakage of their training data, which can be highly sensitive, for example, in medical imaging. Privacy-enhancing technologies, such as diferential privacy (DP), aim to circumvent these susceptibilities. DP is the strongest possible protection for training models while bounding the risks of inferring the inclusion of training samples or reconstructing the original data. DP achieves this by setting a quantifable privacy budget. Although a lower budget decreases the risk of information leakage, it typically also reduces the performance of such models. This imposes a trade-of between robust performance and stringent privacy. Additionally, the interpretation of a privacy budget remains abstract and challenging to contextualize. Here we contrast the performance of artifcial intelligence models at various privacy budgets against both theoretical risk bounds and empirical success of reconstruction attacks. We show that using very large privacy budgets can render reconstruction attacks impossible, while drops in performance are negligible. We thus conclude that not using DP at all is negligent when applying artifcial intelligence models to sensitive data. We deem our results
to lay a foundation for further debates on striking a balance between privacy risks and model performance.
了解更多
