Average-Case Averages: Private Algorithms for Smooth Sensitivity and Mean Estimation
Authors: Mark Bun, Thomas Steinke
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We provide theoretical and experimental evidence showing that our noise distributions compare favorably to others in the literature, in particular, when applied to the mean estimation problem. |
| Researcher Affiliation | Collaboration | Mark Bun Boston University mbun@bu.edu Thomas Steinke IBM Research Almaden smooth@thomas-steinke.net |
| Pseudocode | No | The paper describes algorithms in text and mathematical formulas but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper refers to a full version on arXiv [6] for details and proofs, but does not provide an explicit statement about the release of source code or a direct link to a code repository for the methodology described. |
| Open Datasets | Yes | Our data is sampled from a standard univariate Gaussian distribution. |
| Dataset Splits | No | The paper states that data is sampled from a standard univariate Gaussian distribution and a truncation interval is applied, but it does not specify explicit train, validation, or test dataset splits or percentages. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., CPU, GPU models, memory, or cloud instances) used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers, that are needed to replicate the experiment. |
| Experiment Setup | Yes | The truncation interval is set conservatively to [a, b] = [ 50, 1050] and the data is truncated before applying the trimmed mean... To provide the fairest possible comparison, we pick a ε value (namely, ε = 1 or ε = 0.2) and then compare (ε, 0)-differential privacy with 1/2ε^2-CDP, (1/2ε^2, 10)-t CDP, and (ε, 10^-6)-differential privacy. Each of these is implied by (ε, 0)-differential privacy and the implication is fairly tight, so intuitively provides a roughly similar level of privacy. Aside from the privacy parameters (ε etc.) and the dataset size (n), we show a range of trimming levels m on the horizontal axis. We numerically optimize the smoothing parameter t. We set the distribution shape parameters to appropriate near-optimal values. |