arxiv: 2604.20274 · v2 · submitted 2026-04-22 · 💻 cs.DB

Recognition: unknown

Estimating Power-Law Exponent with Edge Differential Privacy

Adam Tan , Mohamed Hefny , Keval Vora

Authors on Pith no claims yet

Pith reviewed 2026-05-09 23:17 UTC · model grok-4.3

classification 💻 cs.DB

keywords power-law exponentedge differential privacydegree distributionsufficient statisticsgraph privacylocal DPparameter estimationtail cutoff

0 comments

The pith

Releasing low-dimensional sufficient statistics under edge DP reduces distortion when estimating power-law exponents compared to releasing noisy degree distributions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a method for estimating the power-law scaling parameter alpha in graphs while preserving edge differential privacy. Traditional approaches release a noisy version of the full degree distribution first and then fit the model, which introduces high distortion. Instead, the authors identify and privatize only the low-dimensional sufficient statistics needed to estimate alpha directly. These statistics support both discrete approximation and likelihood-based numerical optimization for parameter recovery. The approach is implemented for both centralized and local DP models and tested on real graph datasets under varying privacy budgets and tail cutoffs.

Core claim

By privatizing only the low-dimensional sufficient statistics required for alpha estimation rather than the full degree distribution, edge-DP methods can avoid high distortion and still support accurate recovery of the power-law exponent through discrete approximation or likelihood optimization in centralized and local settings.

What carries the argument

Low-dimensional sufficient statistics for the power-law exponent alpha, released via edge-DP mechanisms such as log-statistic release in the local model.

If this is right

Alpha estimation remains feasible under edge privacy without releasing the entire degree sequence.
Both discrete and continuous optimization methods can recover alpha from the released statistics.
Local DP variants allow comparison between degree release and log-statistic release with lower noise impact.
Performance holds across multiple real-world graph datasets and ranges of privacy budgets.
Tail-cutoff choices interact with the privacy mechanism but still permit consistent estimation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The technique may extend to other graph summary statistics that admit low-dimensional sufficient statistics under DP constraints.
In local DP settings, the reduced noise could enable deployment on resource-limited devices for on-device graph analysis.
Synthetic power-law graphs with controlled alpha could be used to quantify exactly how much the distortion reduction scales with graph size.
If the assumption on sufficient statistics holds, similar strategies might apply to estimating other heavy-tailed parameters like in network traffic or citation data.

Load-bearing premise

The low-dimensional sufficient statistics for alpha can be identified and privatized under edge DP with substantially lower distortion than the full degree distribution while still allowing accurate recovery of alpha via the proposed estimators.

What would settle it

On a graph with known true alpha, compare the estimation error when using the privatized sufficient statistics versus a noisy full degree distribution under identical privacy budget and tail cutoff; if the errors are not meaningfully lower for the sufficient-statistics method, the core claim fails.

Figures

Figures reproduced from arXiv: 2604.20274 by Adam Tan, Keval Vora, Mohamed Hefny.

**Figure 2.** Figure 2: Performance of local DP algorithms. 𝑙1 errors across different combinations for MLE (discrete approximation versus numerical optimization) and local statistic release (degree versus log statistic). Increasing 𝑑min worsens performance for BASE on every dataset and for NO on most datasets; DA, in contrast, improves at 𝑑min = 3 on every dataset. 5.2 Local Algorithms To answer RQ2 and RQ3 for 𝛼ˆLOCAL estimates… view at source ↗

**Figure 3.** Figure 3: Syn-power-1: 𝑙1 vs 𝜀. 0.000005 0.000203 0.000401 0.1 2.0 5.0 0.025877 0.025946 0.026015 l 1 E rro r (a) Centralized. 0.1 2.0 5.0 0.16 0.22 0.29 0.36 0.42 0.49 l 1 E rro r (b) Local [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

read the original abstract

Many real-world graphs have degree distributions that are well approximated by a power-law, and the corresponding scaling parameter $\alpha$ provides a compact summary of that structure which is useful for graph analysis and system optimization. When graphs contain sensitive relationship data, $\alpha$ must be estimated without revealing information about individual edges. This paper studies power-law exponent estimation under edge differential privacy. Instead of first releasing a noisy degree distribution and then fitting a power-law model, we propose privatizing only the low-dimensional sufficient statistics needed to estimate $\alpha$, thereby avoiding the high distortion introduced by traditional approaches. Using these released statistics, we support both discrete approximation and likelihood-based numerical optimization for efficient parameter estimation. We develop edge-DP algorithms for both centralized and local DP models, compare degree release and log-statistic release in the local setting, and evaluate the resulting methods on various graph datasets across multiple privacy budgets and tail-cutoff settings.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Releasing noisy sufficient statistics for the power-law exponent under edge DP gives lower distortion than releasing a full noisy degree histogram, and the paper works out the mechanisms and some experiments.

read the letter

The core move here is to stop releasing a noisy version of the entire degree sequence and instead add noise only to the small set of numbers needed to estimate alpha: the tail count and the sum of log-degrees above a fixed cutoff. Because at most a constant number of edges can affect those two quantities, the Laplace scale stays O(1/epsilon) no matter how large the graph is. That is the main technical point, and it is cleaner than the usual histogram approach whose noise grows with the number of bins.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes estimating the power-law exponent α of graph degree distributions under edge differential privacy by releasing noisy low-dimensional sufficient statistics (tail node count and sum of log-degrees above a fixed cutoff) rather than the full noisy degree histogram. It develops Laplace-based mechanisms for both centralized and local DP, supports discrete approximation and likelihood-based estimators, and reports empirical comparisons on real graphs across privacy budgets and cutoffs.

Significance. If the central claims hold, the work provides a practical utility improvement for a common graph-analytic task by exploiting bounded sensitivity of the chosen statistics, yielding noise scale independent of n. The explicit comparison of degree-histogram release versus log-statistic release in the local model, together with both centralized and local variants, adds concrete guidance for practitioners.

major comments (2)

[§4.1] §4.1 (sensitivity analysis): the claim that the log-degree-sum statistic has sensitivity bounded by a small constant independent of n requires an explicit case analysis for the two possible edge-addition/removal scenarios; without it the O(1/ε) noise-scale advantage over histogram release remains informal.
[§5.2] §5.2 (empirical evaluation): the reported MSE improvements are shown only for a single fixed cutoff per dataset; the paper should include a sensitivity plot or table varying the cutoff to confirm that the advantage persists when the cutoff itself must be chosen from the data.

minor comments (2)

Notation for the tail cutoff parameter is introduced as k in the abstract but appears as τ in the algorithm pseudocode; a single symbol should be used consistently.
[§4.3] The local-DP mechanism description in §4.3 omits the per-user communication cost; adding a short complexity paragraph would clarify the practical difference from the centralized setting.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive review and the constructive comments. We address each major comment below and will incorporate the suggested revisions into the manuscript.

read point-by-point responses

Referee: [§4.1] §4.1 (sensitivity analysis): the claim that the log-degree-sum statistic has sensitivity bounded by a small constant independent of n requires an explicit case analysis for the two possible edge-addition/removal scenarios; without it the O(1/ε) noise-scale advantage over histogram release remains informal.

Authors: We agree that an explicit case analysis is needed to rigorously establish the sensitivity bound. In the revised manuscript, we will expand §4.1 to include a detailed breakdown of the edge-addition and edge-removal scenarios. For each case, we will enumerate the subcases based on whether the two affected nodes lie above or below the fixed cutoff before and after the edge change, and bound the resulting change in the log-degree-sum statistic. This will confirm that the sensitivity is bounded by a small constant that depends only on the cutoff (and is independent of n), thereby making the O(1/ε) noise-scale advantage over full-histogram release formal. revision: yes
Referee: [§5.2] §5.2 (empirical evaluation): the reported MSE improvements are shown only for a single fixed cutoff per dataset; the paper should include a sensitivity plot or table varying the cutoff to confirm that the advantage persists when the cutoff itself must be chosen from the data.

Authors: We appreciate the suggestion to demonstrate robustness with respect to cutoff choice. While the manuscript already reports results across multiple tail-cutoff settings (as stated in the abstract), the primary MSE tables use one representative cutoff per dataset. In the revision we will add a plot (or table) in §5.2 that varies the cutoff over a range for each dataset and shows that the MSE advantage of log-statistic release over histogram release is preserved across these values. This directly addresses the concern about data-dependent cutoff selection. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's core contribution is a methodological choice to release only low-dimensional sufficient statistics (e.g., tail counts and log-degree sums for a fixed cutoff) under edge DP rather than the full noisy degree histogram, followed by standard discrete or likelihood-based estimation of α. These statistics have bounded sensitivity (at most constant, independent of n), allowing Laplace noise with scale O(1/ε) via established DP primitives. Estimation then applies well-known maximum-likelihood or approximation techniques to the released values. No equation or claim reduces the estimator output to a redefinition of the privatized inputs by construction, no uniqueness theorem is imported via self-citation, and no ansatz is smuggled in. The derivation chain is self-contained against external DP mechanisms and statistical estimation literature, with no load-bearing self-referential steps.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

Abstract-only review limits visibility into exact assumptions; standard DP axioms and power-law tail modeling are presumed but not detailed.

free parameters (2)

privacy budget ε
Standard DP parameter varied across experiments; controls noise magnitude.
tail cutoff
Used in evaluations to truncate power-law tails; affects estimator behavior.

axioms (1)

domain assumption Edge differential privacy definition holds for the chosen mechanisms
Invoked implicitly when claiming privacy guarantees for the released statistics.

pith-pipeline@v0.9.0 · 5449 in / 1295 out tokens · 37057 ms · 2026-05-09T23:17:21.092274+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

43 extracted references · 1 canonical work pages

[1]

Fast Exact Shortest-path Dis- tance Queries on Large Networks by Pruned Landmark Labeling

Takuya Akiba, Yoichi Iwata, and Yuichi Yoshida. Fast Exact Shortest-path Dis- tance Queries on Large Networks by Pruned Landmark Labeling. InProceedings of the ACM SIGMOD International Conference on Management of Data, SIG- MOD ’13, page 349–360, 2013

2013
[2]

Engineering Uniform Sampling of Graphs with a Prescribed Power- law Degree Sequence

Daniel Allendorf, Ulrich Meyer, Manuel Penschuck, Hung Tran, and Nick Wormald. Engineering Uniform Sampling of Graphs with a Prescribed Power- law Degree Sequence. In2022 Proceedings of the Symposium on Algorithm Engineering and Experiments (ALENEX), pages 27–40. SIAM, 2022

2022
[3]

Differentially Private Inference via Noisy Optimization.The Annals of Statistics, 51(5):2067 – 2092, 2023

Marco Avella-Medina, Casey Bradshaw, and Po-Ling Loh. Differentially Private Inference via Noisy Optimization.The Annals of Statistics, 51(5):2067 – 2092, 2023

2067
[4]

Parameter Estimation for Power-law Distributions by Maximum Likelihood Methods.The European Physical Journal B, 58(2):167–173, July 2007

Heiko Bauke. Parameter Estimation for Power-law Distributions by Maximum Likelihood Methods.The European Physical Journal B, 58(2):167–173, July 2007

2007
[5]

Kamalika Chaudhuri, Claire Monteleoni, and Anand D. Sarwate. Differentially Private Empirical Risk Minimization.Journal of Machine Learning Research, 12(29):1069–1109, 2011

2011
[6]

Recursive Mechanism: Towards Node Differential Privacy and Unrestricted Joins

Shixi Chen and Shuigeng Zhou. Recursive Mechanism: Towards Node Differential Privacy and Unrestricted Joins. InProceedings of the ACM SIGMOD International Conference on Management of Data, SIGMOD ’13, page 653–664, 2013

2013
[7]

Connected Components in Random Graphs with Given Expected Degree Sequences.Annals of combinatorics, 6(2):125–145, 2002

Fan Chung and Linyuan Lu. Connected Components in Random Graphs with Given Expected Degree Sequences.Annals of combinatorics, 6(2):125–145, 2002

2002
[8]

Power-Law Distri- butions in Empirical Data.SIAM Review, 51(4):661–703, November 2009

Aaron Clauset, Cosma Rohilla Shalizi, and Mark EJ Newman. Power-Law Distri- butions in Empirical Data.SIAM Review, 51(4):661–703, November 2009

2009
[9]

Publishing Graph Degree Distribution with Node Differential Privacy

Wei-Yen Day, Ninghui Li, and Min Lyu. Publishing Graph Degree Distribution with Node Differential Privacy. InProceedings of the International Conference on Management of Data, SIGMOD ’16, page 123–138, 2016

2016
[10]

Liu, Sofya Raskhodnikova, Jessica Shi, Julian Shun, and Shangdi Yu

Laxman Dhulipala, Quanquan C. Liu, Sofya Raskhodnikova, Jessica Shi, Julian Shun, and Shangdi Yu. Differential Privacy from Locally Adjustable Graph Algorithms: k-Core Decomposition, Low Out-Degree Ordering, and Densest Subgraphs. InIEEE 63rd Annual Symposium on F oundations of Computer Science (FOCS), pages 754–765, 2022

2022
[11]

The Algorithmic Foundations of Differential Pri- vacy.F oundations and Trends in Theoretical Computer Science, 9(3–4):211–407, August 2014

Cynthia Dwork and Aaron Roth. The Algorithmic Foundations of Differential Pri- vacy.F oundations and Trends in Theoretical Computer Science, 9(3–4):211–407, August 2014

2014
[12]

Triangle Counting with Local Edge Differential Privacy.Random Structures & Algorithms, 66(4):e70002, 2025

Talya Eden, Quanquan C Liu, Sofya Raskhodnikova, and Adam Smith. Triangle Counting with Local Edge Differential Privacy.Random Structures & Algorithms, 66(4):e70002, 2025

2025
[13]

Accurate Estimation of the Degree Distribution of Private Networks

Michael Hay, Chao Li, Gerome Miklau, and David Jensen. Accurate Estimation of the Degree Distribution of Private Networks. InNinth IEEE International Conference on Data Mining, pages 169–178, 2009

2009
[14]

N2E: A General Framework to Reduce Node- Differential Privacy to Edge-Differential Privacy for Graph Analytics.Proceedings of the ACM on Management of Data, 3(6), December 2025

Yihua Hu, Hao Ding, and Wei Dong. N2E: A General Framework to Reduce Node- Differential Privacy to Edge-Differential Privacy for Graph Analytics.Proceedings of the ACM on Management of Data, 3(6), December 2025

2025
[15]

Locally Differen- tially Private Analysis of Graph Statistics

Jacob Imola, Takao Murakami, and Kamalika Chaudhuri. Locally Differen- tially Private Analysis of Graph Statistics. In30th USENIX Security Symposium (USENIX Security 21), pages 983–1000. USENIX Association, August 2021

2021
[16]

Communication- Efficient Triangle Counting under Local Differential Privacy

Jacob Imola, Takao Murakami, and Kamalika Chaudhuri. Communication- Efficient Triangle Counting under Local Differential Privacy. In31st USENIX Security Symposium (USENIX Security 22), pages 537–554, Boston, MA, August
[17]

Hop Doubling Label Indexing for Point-to-Point Distance Querying on Scale-Free Networks.Proceedings of the VLDB Endowment, 7(12), 2014

Minhao Jiang, Ada Wai-Chee Fu, Raymond Chi-Wing Wong, and Yanyan Xu. Hop Doubling Label Indexing for Point-to-Point Distance Querying on Scale-Free Networks.Proceedings of the VLDB Endowment, 7(12), 2014

2014
[18]

Extremal Mechanisms for Local Differential Privacy.Advances in Neural Information Processing Systems, 27, 2014

Peter Kairouz, Sewoong Oh, and Pramod Viswanath. Extremal Mechanisms for Local Differential Privacy.Advances in Neural Information Processing Systems, 27, 2014

2014
[19]

Private Analysis of Graph Structure.Proceedings of the VLDB Endowment, 4(11):1146–1157, August 2011

Vishesh Karwa, Sofya Raskhodnikova, Adam Smith, and Grigory Yaroslavtsev. Private Analysis of Graph Structure.Proceedings of the VLDB Endowment, 4(11):1146–1157, August 2011

2011
[20]

Analyzing Graphs with Node Differential Privacy

Shiva Prasad Kasiviswanathan, Kobbi Nissim, Sofya Raskhodnikova, and Adam Smith. Analyzing Graphs with Node Differential Privacy. InTheory of Cryptogra- phy, pages 457–476, Berlin, Heidelberg, 2013

2013
[21]

SNAP Datasets: Stanford Large Network Dataset Collection

Jure Leskovec and Andrej Krevl. SNAP Datasets: Stanford Large Network Dataset Collection. http://snap.stanford.edu/data, June 2014

2014
[22]

An Experimental Study on Hub Labeling Based Shortest Path Algorithms.Proceedings of the VLDB Endowment, 11(4):445–457, 2017

Ye Li, Leong Hou U, Man Lung Yiu, and Ngai Meng Kou. An Experimental Study on Hub Labeling Based Shortest Path Algorithms.Proceedings of the VLDB Endowment, 11(4):445–457, 2017

2017
[23]

Dy- namic Graph Publication With Differential Privacy Guarantees for Decentralized Applications.IEEE Transactions on Computers, 74(5):1771–1785, 2025

Zhetao Li, Yong Xiao, Haolin Liu, Xiaofei Liao, Ye Yuan, and Junzhao Du. Dy- namic Graph Publication With Differential Privacy Guarantees for Decentralized Applications.IEEE Transactions on Computers, 74(5):1771–1785, 2025

2025
[24]

Publishing Node Strength Distribution With Node Differential Privacy.IEEE Access, 8:217642–217650, 2020

Ganghong Liu, Xuebin Ma, and Wuyungerile Li. Publishing Node Strength Distribution With Node Differential Privacy.IEEE Access, 8:217642–217650, 2020

2020
[25]

A Crypto- Assisted Approach for Publishing Graph Statistics with Node Local Differential Privacy

Shang Liu, Yang Cao, Takao Murakami, and Masatoshi Yoshikawa. A Crypto- Assisted Approach for Publishing Graph Statistics with Node Local Differential Privacy. InIEEE International Conference on Big Data (Big Data), Osaka, Japan, December 17–20, 2022, pages 5765–5774, 2022

2022
[26]

Mechanism Design via Differential Privacy

Frank McSherry and Kunal Talwar. Mechanism Design via Differential Privacy. InAnnual IEEE Symposium on F oundations of Computer Science (FOCS). IEEE, October 2007

2007
[27]

A Brief History of Generative Models for Power Law and Lognormal Distributions.Internet mathematics, 1(2):226–251, 2004

Michael Mitzenmacher. A Brief History of Generative Models for Power Law and Lognormal Distributions.Internet mathematics, 1(2):226–251, 2004

2004
[28]

Pranay Mundra and Quanquan C. Liu. mundrapranay/DistributedLEDPGraphAl- gos: VLDB Artifact Release, Zenodo , June 2025

2025
[29]

Practical and Accurate Local Edge Differentially Private Graph Algorithms.Pro- ceedings of the VLDB Endowment, 18(11):4199–4213, 2025

Pranay Mundra, Charalampos Papamanthou, Julian Shun, and Quanquan C Liu. Practical and Accurate Local Edge Differentially Private Graph Algorithms.Pro- ceedings of the VLDB Endowment, 18(11):4199–4213, 2025

2025
[30]

Maximum Likelihood Estima- tion of Power-law Degree Distributions via Friendship Paradox-based Sampling

Buddhika Nettasinghe and Vikram Krishnamurthy. Maximum Likelihood Estima- tion of Power-law Degree Distributions via Friendship Paradox-based Sampling. ACM Transactions on Knowledge Discovery from Data (TKDD), 15(6), May 2021

2021
[31]

M. E. J. Newman. Power Laws, Pareto Distributions and Zipf’s Law.Contempo- rary Physics, 46(5):323–351, 2005

2005
[32]

Gener- ating Synthetic Decentralized Social Graphs with Local Differential Privacy

Zhan Qin, Ting Yu, Yin Yang, Issa Khalil, Xiaokui Xiao, and Kui Ren. Gener- ating Synthetic Decentralized Social Graphs with Local Differential Privacy. In Proceedings of the ACM SIGSAC Conference on Computer and Communications Security, CCS ’17, page 425–438, 2017

2017
[33]

Towards a Differential Privacy Theory for Edge-Labeled Directed Graphs

Jenni Reuben. Towards a Differential Privacy Theory for Edge-Labeled Directed Graphs. InSICHERHEIT 2018, pages 273–278. Gesellschaft für Informatik e.V ., Bonn, 2018

2018
[34]

Optimizing and Auto-tuning Scale-free Sparse Matrix-vector Multiplication on Intel Xeon Phi

Wai Teng Tang, Ruizhe Zhao, Mian Lu, Yun Liang, Huynh Phung Huyng, Xibai Li, and Rick Siow Mong Goh. Optimizing and Auto-tuning Scale-free Sparse Matrix-vector Multiplication on Intel Xeon Phi. InIEEE/ACM International Symposium on Code Generation and Optimization (CGO), pages 136–145. IEEE, 2015

2015
[35]

Efficiently Estimating Erdos-Renyi Graphs with Node Differential Privacy

Jonathan Ullman and Adam Sealfon. Efficiently Estimating Erdos-Renyi Graphs with Node Differential Privacy. In H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché-Buc, E. Fox, and R. Garnett, editors,Advances in Neural Information Processing Systems, volume 32. Curran Associates, Inc., 2019

2019
[36]

Lumos: Dependency-Driven Disk-based Graph Processing

Keval V ora. Lumos: Dependency-Driven Disk-based Graph Processing. In USENIX Annual Technical Conference (USENIX ATC 19), pages 429–442, 2019

2019
[37]

Songlei Wang, Yifeng Zheng, Xiaohua Jia, and Haibo Hu. PrivAGM: Secure Construction of Differentially Private Directed Attributed Graph Models on Decen- tralized Social Graphs.Proceedings of the VLDB Endowment, 18(11):4682–4694, July 2025

2025
[38]

Preserving Differential Privacy in Degree-Correlation based Graph Generation.Transactions on Data Privacy, 6(2):127–145, August 2013

Yue Wang and Xintao Wu. Preserving Differential Privacy in Degree-Correlation based Graph Generation.Transactions on Data Privacy, 6(2):127–145, August 2013

2013
[39]

AsgLDP: Collecting and Generating Decentralized Attributed Graphs With Local Differential Privacy.IEEE Transactions on Information F orensics and Security, 15:3239–3254, 2020

Chengkun Wei, Shouling Ji, Changchang Liu, Wenzhi Chen, and Ting Wang. AsgLDP: Collecting and Generating Decentralized Attributed Graphs With Local Differential Privacy.IEEE Transactions on Information F orensics and Security, 15:3239–3254, 2020

2020
[40]

Sectric: Towards Accurate, Privacy- Preserving and Efficient Triangle Counting.Proceedings of the VLDB Endowment, 18(10):3382–3395, June 2025

Minze Xu, Zhentai Xie, Zhibin Wang, Guangzhan Wang, Longbin Lai, Yuan Zhang, Chen Tian, and Sheng Zhong. Sectric: Towards Accurate, Privacy- Preserving and Efficient Triangle Counting.Proceedings of the VLDB Endowment, 18(10):3382–3395, June 2025

2025
[41]

PrivGraph: Differentially Private Graph Data Publication by Exploiting Community Information

Quan Yuan, Zhikun Zhang, Linkang Du, Min Chen, Peng Cheng, and Mingyang Sun. PrivGraph: Differentially Private Graph Data Publication by Exploiting Community Information. In32nd USENIX Security Symposium (USENIX Security 23), pages 3241–3258, 2023

2023
[42]

PrivDPR: Synthetic Graph Publishing with Deep PageRank under Differential Privacy

Sen Zhang, Haibo Hu, Qingqing Ye, and Jianliang Xu. PrivDPR: Synthetic Graph Publishing with Deep PageRank under Differential Privacy. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V .1, KDD ’25, page 1936–1947, 2025

1936
[43]

A Two-Phase Algorithm for Generating Synthetic Graph Under Local Differential Privacy

Yuxuan Zhang, Jianghong Wei, Xiaojian Zhang, Xuexian Hu, and Wenfen Liu. A Two-Phase Algorithm for Generating Synthetic Graph Under Local Differential Privacy. InProceedings of the 8th International Conference on Communication and Network Security, ICCNS ’18, page 84–89, 2018. 9 SeQureDB ’26, May 31-June 05, 2026, Bengaluru, India Adam T an, Mohamed Hefny...

work page arXiv 2018