The recently proposed Fast Fourier Transform (FFT)-based accountant for evaluating $(\varepsilon,\delta)$-differential privacy guarantees using the privacy loss distribution formalism has been shown to give tighter bounds than commonly used methods such as R\'enyi accountants when applied to compositions of homogeneous mechanisms. This approach is also applicable to certain discrete mechanisms that cannot be analysed with R\'enyi accountants. In this paper, we extend this approach to compositions of heterogeneous mechanisms. We carry out a full error analysis that allows choosing the parameters of the algorithm such that a desired accuracy is obtained. Using our analysis, we also give a bound for the computational complexity in terms of the error which is analogous to and slightly tightens the one given by Murtagh and Vadhan (2018). We also show how to speed up the evaluation of tight privacy guarantees using the Plancherel theorem at the cost of increased pre-computation and memory usage.
翻译:最近提议的Fast Fourier变换(FFT)会计在评估美元(varepsilon,\delta)美元(varepsilon,\delta)的基础上使用隐私损失分配形式,对使用隐私损失分配格式的隐私保障进行区分,这与R\'enyi会计师等常用方法应用到同质机制的构成时相比,显示出了更严格的限制。这个方法也适用于某些无法与R\'enyi会计师分析的离散机制。在本文中,我们将这一方法扩大到多种机制的构成。我们进行了全面的错误分析,以便选择算法的参数,从而达到预期的准确性。我们的分析还结合了类似于Murtagh和Vadhan(2018年)给出的错误的计算复杂性,并略微收紧了这两个错误。我们还展示了如何加快对使用Plancherel理论的严格隐私保障的评估,以增加计算前和记忆使用的成本。