Version 1
: Received: 4 June 2023 / Approved: 5 June 2023 / Online: 5 June 2023 (09:36:31 CEST)
Version 2
: Received: 5 June 2023 / Approved: 8 June 2023 / Online: 8 June 2023 (03:16:47 CEST)
How to cite:
Li, X.; Zhu, H.; Zhang, Z.; Xie, M. Item-Oriented Personalized LDP for Discrete Distribution Estimation. Preprints2023, 2023060289. https://doi.org/10.20944/preprints202306.0289.v2
Li, X.; Zhu, H.; Zhang, Z.; Xie, M. Item-Oriented Personalized LDP for Discrete Distribution Estimation. Preprints 2023, 2023060289. https://doi.org/10.20944/preprints202306.0289.v2
Li, X.; Zhu, H.; Zhang, Z.; Xie, M. Item-Oriented Personalized LDP for Discrete Distribution Estimation. Preprints2023, 2023060289. https://doi.org/10.20944/preprints202306.0289.v2
APA Style
Li, X., Zhu, H., Zhang, Z., & Xie, M. (2023). Item-Oriented Personalized LDP for Discrete Distribution Estimation. Preprints. https://doi.org/10.20944/preprints202306.0289.v2
Chicago/Turabian Style
Li, X., Zhiqiang Zhang and Meiyi Xie. 2023 "Item-Oriented Personalized LDP for Discrete Distribution Estimation" Preprints. https://doi.org/10.20944/preprints202306.0289.v2
Abstract
Discrete distribution estimation is a fundamental statistical tool, which is widely used to perform data analysis tasks in various applications involving sensitive personal information. Due to privacy concerns, individuals may not always provide their raw information, which leads to unpredictable biases in the final results of estimated distribution. Local Differential Privacy (LDP) is an advanced technique for privacy protection of discrete distribution estimation. Currently, typical LDP mechanisms provide same protection for all items in the domain, which imposes unnecessary perturbation on less sensitive items and thus degrades the utility of final results. Although, several recent works try to alleviate this problem, the utility can be further improved. In this paper, we propose a novel notion called Item-Oriented Personalized LDP (IPLDP), which independently perturbs different items with different privacy budgets to achieve personalized privacy protection. Furthermore, to satisfy IPLDP, we propose the Item-Oriented Personalized Randomized Response (IPRR) based on the observation that the sensitivity of data shows an inverse relationship with the population size of respective individuals. Theoretical analysis and experimental results demonstrate that our method can provide fine-grained privacy protection and improve data utility simultaneously.
Keywords
Discrete distribution estimation; Local differential privacy; Item-oriented personalization; Randomized response
Subject
Computer Science and Mathematics, Other
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Received:
8 June 2023
Commenter:
Xin Li
Commenter's Conflict of Interests:
Author
Comment:
We have made some changes to the manuscript to correct a mistake that was identified after the initial submission. Specifically, we changed the equations and added a sentence of explanation after the changed equations in B.3 on the top half of page 20.
Commenter: Xin Li
Commenter's Conflict of Interests: Author