Consistent Estimation of Numerical Distributions under Local Differential Privacy by Wavelet Expansion

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Puning, Zhang, Zhikun, Sun, Bo, Shen, Li, Zhang, Liang, Wang, Shaowei, Liu, Zhe
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909803905810432
author Zhao, Puning
Zhang, Zhikun
Sun, Bo
Shen, Li
Zhang, Liang
Wang, Shaowei
Liu, Zhe
author_facet Zhao, Puning
Zhang, Zhikun
Sun, Bo
Shen, Li
Zhang, Liang
Wang, Shaowei
Liu, Zhe
contents Distribution estimation under local differential privacy (LDP) is a fundamental and challenging task. Significant progresses have been made on categorical data. However, due to different evaluation metrics, these methods do not work well when transferred to numerical data. In particular, we need to prevent the probability mass from being misplaced far away. In this paper, we propose a new approach that express the sample distribution using wavelet expansions. The coefficients of wavelet series are estimated under LDP. Our method prioritizes the estimation of low-order coefficients, in order to ensure accurate estimation at macroscopic level. Therefore, the probability mass is prevented from being misplaced too far away from its ground truth. We establish theoretical guarantees for our methods. Experiments show that our wavelet expansion method significantly outperforms existing solutions under Wasserstein and KS distances.
format Preprint
id arxiv_https___arxiv_org_abs_2509_19661
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Consistent Estimation of Numerical Distributions under Local Differential Privacy by Wavelet Expansion
Zhao, Puning
Zhang, Zhikun
Sun, Bo
Shen, Li
Zhang, Liang
Wang, Shaowei
Liu, Zhe
Machine Learning
Distribution estimation under local differential privacy (LDP) is a fundamental and challenging task. Significant progresses have been made on categorical data. However, due to different evaluation metrics, these methods do not work well when transferred to numerical data. In particular, we need to prevent the probability mass from being misplaced far away. In this paper, we propose a new approach that express the sample distribution using wavelet expansions. The coefficients of wavelet series are estimated under LDP. Our method prioritizes the estimation of low-order coefficients, in order to ensure accurate estimation at macroscopic level. Therefore, the probability mass is prevented from being misplaced too far away from its ground truth. We establish theoretical guarantees for our methods. Experiments show that our wavelet expansion method significantly outperforms existing solutions under Wasserstein and KS distances.
title Consistent Estimation of Numerical Distributions under Local Differential Privacy by Wavelet Expansion
topic Machine Learning
url https://arxiv.org/abs/2509.19661