Context Aware Local Differential Privacy
Authors: Jayadev Acharya, Kallista Bonawitz, Peter Kairouz, Daniel Ramage, Ziteng Sun
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform experiments on both synthetic data and real data to empirically validate how the new notion of context-aware LDP and associated algorithms would affect the accuracy of k-ary distribution estimation. |
| Researcher Affiliation | Collaboration | 1ECE, Cornell University, Ithaca, New York, USA 2Google, Seattle, USA |
| Pseudocode | No | The paper describes algorithms and derivations (e.g., in Sections 5.1 and 6.1) through mathematical formulas and descriptive text, but it does not include formal pseudocode blocks or algorithms labeled as such. |
| Open Source Code | No | The paper does not provide any explicit statement about making the source code open, nor does it include a link to a code repository for the described methodology. |
| Open Datasets | Yes | To validate our algorithm on real datasets, we take the Gowalla user check-in dataset (Cho et al., 2011) |
| Dataset Splits | No | The paper does not describe explicit train/validation/test dataset splits. It discusses generating samples or using existing datasets for distribution estimation, but not in terms of partitioning data for model training and validation. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers that would be required to reproduce the experiments. |
| Experiment Setup | Yes | For synthetic data, we set k = 1000, " = 1. We assume all the blocks to have the same size and m 2 {10, 20, 50, 100}. We take n = 1000 2i, i 2 {0, 1, , 9}. We partition latitudes into m1 equal parts and longitudes into m2 equal parts. The resulting grid will be used as the blocks (m1m2 blocks in total). Table 1 shows the average d T V error over 100 runs of the experiment for LDP and BS-LDP with different (m1, m2) pairs. |