The advancement of technology and the widespread usage of smart phones have made the collection of data from users easy and cost-effective, which allows the government, urban planner, and researchers to envision novel analysis. Along with the benefits, the shared data can bring serious privacy concerns as they reveal sensitive information about a user. Differential privacy has become an effective model for sharing privacy protected data with others. To facilitate users to protect the privacy of data before it leaves their personal devices, the concept of personal local differential privacy (PLDP) has been introduced for counting queries. We formulate PLDP for computing aggregates over numeric data. We present an efficient approach, private estimation of numeric aggregates (PENA), that guarantees PLDP of numeric data while computing an aggregate (e.g., the average or the minimum). We perform extensive experiments over a real dataset to show the effectiveness of PENA.
- This repository contains the titled paper and code
For citation:
@inproceedings{akter2017computing,
title={Computing aggregates over numeric data with personalized local differential privacy},
author={Akter, Mousumi and Hashem, Tanzima},
booktitle={Australasian Conference on Information Security and Privacy},
pages={249--260},
year={2017},
organization={Springer}
}