Decision Making with Differential Privacy under a Fairness Lens

Authors: Cuong Tran, Ferdinando Fioretto, Pascal Van Hentenryck, Zhiyan Yao

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The proposed approaches are evaluated on critical decision problems that use differentially private census data.As illustrated in Figure 2, Problem PF can introduce significant disparity errors. For ϵ = 0.001, 0.01, and 0.1 the estimated fairness bounds are 0.003, 3 10 5, and 1.2 10 6 respectively, which amount to an average misallocation of $43,281, $4,328, and $865.6 respectively.
Researcher Affiliation Academia 1Syracuse University 2Georgia Institute of Technology 3Nanjing University of Science and Technology {cutran, fiorett}@syr.edu, pvh@isye.gatech.edu, zyao09@syr.edu
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain an explicit statement about releasing source code for the methodology described, nor does it provide a link to a code repository.
Open Datasets No The paper mentions using 'differentially private census data' and refers to real-world applications based on census data, such as 'NYC census data' in figures. However, it does not provide a specific link, DOI, repository, or formal citation (with authors and year) to the exact dataset used in their experiments.
Dataset Splits No The paper does not provide specific details about training, validation, or test dataset splits.
Hardware Specification No The paper does not specify the hardware (e.g., GPU/CPU models, memory) used to run its experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, or specialized solvers) used in the experiments.
Experiment Setup Yes For ϵ = 0.001, 0.01, and 0.1 the estimated fairness bounds are 0.003, 3 10 5, and 1.2 10 6 respectively...The estimated fairness bounds were obtained by performing a linear search over all n school districts and selecting the maximal Tr(HPF i ).obtained using m = 104 repetitions.Here T is a temperature parameter that controls the strengths of the correction.