Calibrating Noise for Group Privacy in Subsampled Mechanisms

Summary: Introduces tight privacy accounting for group privacy in subsampled mechanisms, avoiding pessimistic black-box DP-to-GP conversion. Exploiting subsampling randomness yields over an order-of-magnitude noise reductions, including for GP-SGD deep-learning training. (summarized by gpt-5.6-luna on Jul 24 2026)

Paper ID: h92f728f9cd9eead7
Venue: VLDB
Year: 2025
Pagerank: 4.9605343e-05
Overall Rank: 11,370 | 23.79%
DOI: 10.14778/3705829.3705848
PDF: Download (CC BY-NC-ND 4.0)

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. Yangfan Jiang (National University of Singapore)
2. Xinjian Luo (National University of Singapore)
3. Yin Yang (Hamad Bin Khalifa University)
4. Xiaokui Xiao (National University of Singapore)

BibTeX Citation

@article{jiang_vldb25,
        title = {{Calibrating Noise for Group Privacy in Subsampled Mechanisms}},
        author = {Jiang, Yangfan and Luo, Xinjian and Yang, Yin and Xiao, Xiaokui},
        journal = {PVLDB},
        series = {{VLDB} '25},
        volume = {18},
        number = {2},
        pages = {322--334},
        doi = {10.14778/3705829.3705848},
        url = {https://doi.org/10.14778/3705829.3705848},
        year = {2025}
}

Incoming Citations (Sorted by Pagerank)

Showing 1 of 1 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
10,518	DP-S4S: Accurate and Scalable Select-Join-Aggregate Query Processing with User-Level Differential Privacy	2026	SIGMOD	4.9605343e-05

Outgoing Citations (Sorted by Pagerank)

Showing 17 of 17 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
64	Privacy Integrated Queries: An Extensible Platform for Privacy-Preserving Data Analysis	2009	SIGMOD	0.00038378358
123	Revealing Information while Preserving Privacy	2003	PODS	0.00030654879
1,170	PrivateSQL: A Differentially Private SQL Query Engine	2019	VLDB	0.00011687052
1,256	No Free Lunch in Data Privacy	2011	SIGMOD	0.00011304781
2,288	DPTree: Differential Indexing for Persistent Memory	2020	VLDB	8.6817411e-05
3,922	Pufferfish Privacy Mechanisms for Correlated Data	2017	SIGMOD	6.9051301e-05
4,518	R2T: Instance-optimal Truncation for Differentially Private Query Evaluation with Foreign Keys	2022	SIGMOD	6.551704e-05
4,558	Projected Federated Averaging with Heterogeneous Differential Privacy	2022	VLDB	6.5249713e-05
5,171	A Rigorous and Customizable Framework for Privacy	2012	PODS	6.2321934e-05
5,307	Functional Mechanism: Regression Analysis under Differential Privacy	2012	VLDB	6.1741542e-05
6,323	Longshot: Indexing Growing Databases using MPC and Differential Privacy	2023	VLDB	5.7958256e-05
6,350	A Neural Approach to Spatio-Temporal Data Release with User-Level Differential Privacy	2023	SIGMOD	5.788067e-05
6,551	Skellam Mixture Mechanism: a Novel Approach to Federated Learning with Differential Privacy	2022	VLDB	5.7316321e-05
6,566	Falcon: A Privacy-Preserving and Interpretable Vertical Federated Learning System	2023	VLDB	5.728236e-05
7,626	Privacy Amplification by Sampling under User-level Differential Privacy	2024	SIGMOD	5.4657865e-05
9,016	Cache Me If You Can: Accuracy-Aware Inference Engine for Differentially Private Data Exploration	2023	VLDB	5.2232014e-05
9,782	Free Gap Information from the Differentially Private Sparse Vector and Noisy Max Mechanisms	2020	VLDB	5.115551e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	10,615	A General Framework for Per-record Differential Privacy	2026	SIGMOD
2	10,803	Understanding Disclosure Risk in Differential Privacy with Applications to Noise Calibration and Auditing	2026	VLDB
3	858	Recursive Mechanism: Towards Node Differential Privacy and Unrestricted Joins	2013	SIGMOD
4	10,711	Enhancing Local Differential Privacy Accuracy by Exploiting Inherent Uncertainty	2026	SIGMOD
5	8,022	Quantifying identifiability to choose and audit epsilon in differentially private deep learning	2021	VLDB
6	10,743	Privacy Loss of Noise Perturbation via Concentration Analysis of A Product Measure	2026	SIGMOD
7	801	Differentially Private Data Cubes: Optimizing Noise Sources and Consistency	2011	SIGMOD
8	7,626	Privacy Amplification by Sampling under User-level Differential Privacy	2024	SIGMOD
9	5,652	Practical Differential Privacy via Grouping and Smoothing	2013	VLDB
10	6,551	Skellam Mixture Mechanism: a Novel Approach to Federated Learning with Differential Privacy	2022	VLDB