Differentially Private Change-Point Detection
Authors: Rachel Cummings, Sara Krehbiel, Yajun Mei, Rui Tuo, Wanrong Zhang
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We study the statistical problem of change-point detection through the lens of differential privacy. We give private algorithms for both online and offline change-point detection, analyze these algorithms theoretically, and provide empirical validation of our results. |
| Researcher Affiliation | Academia | Rachel Cummings Georgia Institute of Technology rachelc@gatech.edu Sara Krehbiel University of Richmond krehbiel@richmond.edu Yajun Mei Georgia Institute of Technology ymei@gatech.edu Rui Tuo Texas A&M University ruituo@tamu.edu Wanrong Zhang Georgia Institute of Technology wanrongz@gatech.edu |
| Pseudocode | Yes | Algorithm 1 Offline private change-point detector : OFFLINEPCPD(X, P0, P1, , δ, n) and Algorithm 2 Online private change-point detector : ONLINEPCPD(X, P0, P1, , n, T) |
| Open Source Code | No | The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | The paper describes generating data for Monte Carlo experiments from Bernoulli and Gaussian distributions, e.g., “We consider data drawn from Bernoulli and Gaussian distributions”. It does not refer to or provide access to a specific publicly available dataset. |
| Dataset Splits | No | The paper describes Monte Carlo simulations and parameters like 'n = 200 observations' and '10^4 times' but does not specify train, validation, or test dataset splits in the conventional sense, as the data is generated for each run rather than partitioned from a fixed dataset. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments (e.g., CPU, GPU models, or memory specifications). |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers for replication (e.g., Python 3.8, PyTorch 1.9). |
| Experiment Setup | Yes | We use n = 200 observations where the true change occurs at time k = 100. This process is repeated 104 times. For both the Bernoulli and Gaussian models, we consider the following three different change scenarios... For each of these cases, we consider privacy parameter = 0.1, 0.5, 1, 1, where = 1 corresponds to the non-private problem, which serves as our baseline. |