Composition theorems for f-differential privacy

Natasha Fernandes; Annabelle McIver; Parastoo Sadeghi

doi:10.1007/978-3-032-22730-0_22

Composition theorems for f-differential privacy

Natasha Fernandes, Annabelle McIver^*, Parastoo Sadeghi

^*Corresponding author for this work

School of Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

Abstract

f-differential privacy (f-DP) is a recent definition for privacy which can offer improved predictions of “privacy loss”. It has been used to analyse specific privacy mechanisms, such as the popular Gaussian mechanism.

In this paper we show how f-DP’s foundation in statistical hypothesis testing implies equivalence to the channel model of Quantitative Information Flow (QIF). We demonstrate this equivalence as a Galois connection between two partially-ordered sets, namely f-DP’s trade-off functions, and a class of information channels. This equivalence enables novel general composition theorems for f-DP, supporting improved analysis for complex privacy designs. We apply our results to the popular privacy amplification mechanisms of sub-sampling and purification, to produce novel f-DP profiles for these general privacy-enhancing algorithms.

Original language	English
Title of host publication	Foundations of Software Science and Computation Structures
Subtitle of host publication	29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings
Editors	Nathalie Bertrand, Stefan Milius
Place of Publication	Cham, Switzerland
Publisher	Springer, Springer Nature
Pages	462-483
Number of pages	22
ISBN (Electronic)	9783032227300
ISBN (Print)	9783032227294
DOIs	https://doi.org/10.1007/978-3-032-22730-0_22
Publication status	Published - 2026
Event	29th International Conference on Foundations of Software Science and Computation Structures, FoSSaCS 2026, Held as Part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026 - Turin, Italy Duration: 11 Apr 2026 → 16 Apr 2026

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	16503
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	29th International Conference on Foundations of Software Science and Computation Structures, FoSSaCS 2026, Held as Part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026
Country/Territory	Italy
City	Turin
Period	11/04/26 → 16/04/26

Bibliographical note

Copyright the Author(s) 2026. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Keywords

Quantitative information flow
semantics for probabilistic programs
compositional analyses for privacy

Access to Document

10.1007/978-3-032-22730-0_22Licence: CC BY-NC-ND

Publisher versionFinal published version, 428 KBLicence: CC BY-NC-ND

Cite this

Fernandes, N., McIver, A., & Sadeghi, P. (2026). Composition theorems for f-differential privacy. In N. Bertrand, & S. Milius (Eds.), Foundations of Software Science and Computation Structures: 29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings (pp. 462-483). (Lecture Notes in Computer Science; Vol. 16503). Springer, Springer Nature. https://doi.org/10.1007/978-3-032-22730-0_22

Fernandes, Natasha ; McIver, Annabelle ; Sadeghi, Parastoo. / Composition theorems for f-differential privacy. Foundations of Software Science and Computation Structures: 29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings. editor / Nathalie Bertrand ; Stefan Milius. Cham, Switzerland : Springer, Springer Nature, 2026. pp. 462-483 (Lecture Notes in Computer Science).

@inproceedings{3c644921c197449d95e69b2a2651fe61,

title = "Composition theorems for f-differential privacy",

abstract = "f-differential privacy (f-DP) is a recent definition for privacy which can offer improved predictions of “privacy loss”. It has been used to analyse specific privacy mechanisms, such as the popular Gaussian mechanism. In this paper we show how f-DP{\textquoteright}s foundation in statistical hypothesis testing implies equivalence to the channel model of Quantitative Information Flow (QIF). We demonstrate this equivalence as a Galois connection between two partially-ordered sets, namely f-DP{\textquoteright}s trade-off functions, and a class of information channels. This equivalence enables novel general composition theorems for f-DP, supporting improved analysis for complex privacy designs. We apply our results to the popular privacy amplification mechanisms of sub-sampling and purification, to produce novel f-DP profiles for these general privacy-enhancing algorithms.",

keywords = "Quantitative information flow, semantics for probabilistic programs, compositional analyses for privacy",

author = "Natasha Fernandes and Annabelle McIver and Parastoo Sadeghi",

note = "Copyright the Author(s) 2026. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.; 29th International Conference on Foundations of Software Science and Computation Structures, FoSSaCS 2026, Held as Part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026 ; Conference date: 11-04-2026 Through 16-04-2026",

year = "2026",

doi = "10.1007/978-3-032-22730-0\_22",

language = "English",

isbn = "9783032227294",

series = "Lecture Notes in Computer Science",

publisher = "Springer, Springer Nature",

pages = "462--483",

editor = "Nathalie Bertrand and Stefan Milius",

booktitle = "Foundations of Software Science and Computation Structures",

address = "United States",

}

Fernandes, N , McIver, A & Sadeghi, P 2026, Composition theorems for f-differential privacy. in N Bertrand & S Milius (eds), Foundations of Software Science and Computation Structures: 29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings. Lecture Notes in Computer Science, vol. 16503, Springer, Springer Nature, Cham, Switzerland, pp. 462-483, 29th International Conference on Foundations of Software Science and Computation Structures, FoSSaCS 2026, Held as Part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, 11/04/26. https://doi.org/10.1007/978-3-032-22730-0_22

Composition theorems for f-differential privacy. / Fernandes, Natasha ; McIver, Annabelle; Sadeghi, Parastoo.
Foundations of Software Science and Computation Structures: 29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings. ed. / Nathalie Bertrand; Stefan Milius. Cham, Switzerland: Springer, Springer Nature, 2026. p. 462-483 (Lecture Notes in Computer Science; Vol. 16503).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Composition theorems for f-differential privacy

AU - Fernandes, Natasha

AU - McIver, Annabelle

AU - Sadeghi, Parastoo

N1 - Copyright the Author(s) 2026. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

PY - 2026

Y1 - 2026

N2 - f-differential privacy (f-DP) is a recent definition for privacy which can offer improved predictions of “privacy loss”. It has been used to analyse specific privacy mechanisms, such as the popular Gaussian mechanism. In this paper we show how f-DP’s foundation in statistical hypothesis testing implies equivalence to the channel model of Quantitative Information Flow (QIF). We demonstrate this equivalence as a Galois connection between two partially-ordered sets, namely f-DP’s trade-off functions, and a class of information channels. This equivalence enables novel general composition theorems for f-DP, supporting improved analysis for complex privacy designs. We apply our results to the popular privacy amplification mechanisms of sub-sampling and purification, to produce novel f-DP profiles for these general privacy-enhancing algorithms.

AB - f-differential privacy (f-DP) is a recent definition for privacy which can offer improved predictions of “privacy loss”. It has been used to analyse specific privacy mechanisms, such as the popular Gaussian mechanism. In this paper we show how f-DP’s foundation in statistical hypothesis testing implies equivalence to the channel model of Quantitative Information Flow (QIF). We demonstrate this equivalence as a Galois connection between two partially-ordered sets, namely f-DP’s trade-off functions, and a class of information channels. This equivalence enables novel general composition theorems for f-DP, supporting improved analysis for complex privacy designs. We apply our results to the popular privacy amplification mechanisms of sub-sampling and purification, to produce novel f-DP profiles for these general privacy-enhancing algorithms.

KW - Quantitative information flow

KW - semantics for probabilistic programs

KW - compositional analyses for privacy

UR - https://www.scopus.com/pages/publications/105039562374

U2 - 10.1007/978-3-032-22730-0_22

DO - 10.1007/978-3-032-22730-0_22

M3 - Conference proceeding contribution

AN - SCOPUS:105039562374

SN - 9783032227294

T3 - Lecture Notes in Computer Science

SP - 462

EP - 483

BT - Foundations of Software Science and Computation Structures

A2 - Bertrand, Nathalie

A2 - Milius, Stefan

PB - Springer, Springer Nature

CY - Cham, Switzerland

T2 - 29th International Conference on Foundations of Software Science and Computation Structures, FoSSaCS 2026, Held as Part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026

Y2 - 11 April 2026 through 16 April 2026

ER -

Fernandes N , McIver A, Sadeghi P. Composition theorems for f-differential privacy. In Bertrand N, Milius S, editors, Foundations of Software Science and Computation Structures: 29th International Conference, FoSSaCS 2026, held as part of the International Joint Conferences on Theory and Practice of Software, ETAPS 2026, Turin, Italy, April 11-16, 2026, proceedings. Cham, Switzerland: Springer, Springer Nature. 2026. p. 462-483. (Lecture Notes in Computer Science). doi: 10.1007/978-3-032-22730-0_22

Composition theorems for f-differential privacy

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this