Instance-optimal Mean Estimation Under Differential Privacy
Authors: Ziyue Huang, Yuting Liang, Ke Yi
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We performed both statistical and empirical mean estimation experiments to evaluate our method. For statistical mean estimation, we used multivariate Gaussian distributions with various µ and Σ. ... For empirical mean estimation, we used a real-world dataset, MNIST... We measured the ℓ2 error by taking the trimmed mean with trimming parameter 0.1 over 100 trials (as in [8]). |
| Researcher Affiliation | Academia | Ziyue Huang, Yuting Liang, Ke Yi {zhuangbq,yliangbs,yike}@cse.ust.hk Department of Computer Science and Engineering Hong Kong University of Science and Technology |
| Pseudocode | No | The paper mentions 'Our algorithm Priv Quant' and refers to 'details in the supplementary material', but no pseudocode or algorithm blocks are present within the provided text of the paper. |
| Open Source Code | No | The paper does not provide any specific links to open-source code repositories or explicit statements about code availability. |
| Open Datasets | Yes | For empirical mean estimation, we used a real-world dataset, MNIST, which consists of 70,000 images of handwritten digits, where each image is represented by a vector of dimension d = 784 = 28 28. ... For statistical mean estimation, we used multivariate Gaussian distributions with various µ and Σ. |
| Dataset Splits | No | The paper mentions using the MNIST dataset and statistical mean estimation with Gaussian distributions, but it does not provide specific details on how the datasets were split into training, validation, or test sets (e.g., percentages, sample counts, or citations to specific splits). |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU or CPU models, memory specifications, or types of computing resources used for the experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers (e.g., Python 3.8, PyTorch 1.9). |
| Experiment Setup | Yes | For statistical mean estimation, we used multivariate Gaussian distributions with various µ and Σ. All algorithms are given the same R, σmin, σmax. We tried various R, while fixing σmin = 0.1 and σmax = R/ d. ... We measured the ℓ2 error by taking the trimmed mean with trimming parameter 0.1 over 100 trials (as in [8]). |