A Tunable Measure for Information Leakage

Jiachun Liao, Oliver Kosut, Lalitha Sankar, Flavio P. Calmon

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

A tunable measure for information leakage called maximal a-leakage is introduced. This measure quantifies the maximal gain of an adversary in refining a tilted version of its prior belief of any (potentially random) function of a dataset conditioning on a disclosed dataset. The choice of \alpha determines the specific adversarial action ranging from refining a belief for \alpha=1 to guessing the best posterior for \alpha=\infty, and for these extremal values this measure simplifies to mutual information (MI) and maximal leakage (MaxL), respectively. For all other \alpha this measure is shown to be the Arimoto channel capacity. Several properties of this measure are proven including: (i) quasi-convexity in the mapping between the original and disclosed datasets; (ii) data processing inequalities; and (iii) a composition property. A full version of this paper is in [1].

Original languageEnglish (US)
Title of host publication2018 IEEE International Symposium on Information Theory, ISIT 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages701-705
Number of pages5
Volume2018-June
ISBN (Print)9781538647806
DOIs
StatePublished - Aug 15 2018
Event2018 IEEE International Symposium on Information Theory, ISIT 2018 - Vail, United States
Duration: Jun 17 2018Jun 22 2018

Other

Other2018 IEEE International Symposium on Information Theory, ISIT 2018
CountryUnited States
CityVail
Period6/17/186/22/18

Fingerprint

Leakage
Refining
Channel capacity
Quasiconvexity
Chemical analysis
Channel Capacity
Random Function
Mutual Information
Conditioning
Simplify
Quantify
Beliefs

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Information Systems
  • Modeling and Simulation
  • Applied Mathematics

Cite this

Liao, J., Kosut, O., Sankar, L., & Calmon, F. P. (2018). A Tunable Measure for Information Leakage. In 2018 IEEE International Symposium on Information Theory, ISIT 2018 (Vol. 2018-June, pp. 701-705). [8437307] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ISIT.2018.8437307

A Tunable Measure for Information Leakage. / Liao, Jiachun; Kosut, Oliver; Sankar, Lalitha; Calmon, Flavio P.

2018 IEEE International Symposium on Information Theory, ISIT 2018. Vol. 2018-June Institute of Electrical and Electronics Engineers Inc., 2018. p. 701-705 8437307.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Liao, J, Kosut, O, Sankar, L & Calmon, FP 2018, A Tunable Measure for Information Leakage. in 2018 IEEE International Symposium on Information Theory, ISIT 2018. vol. 2018-June, 8437307, Institute of Electrical and Electronics Engineers Inc., pp. 701-705, 2018 IEEE International Symposium on Information Theory, ISIT 2018, Vail, United States, 6/17/18. https://doi.org/10.1109/ISIT.2018.8437307
Liao J, Kosut O, Sankar L, Calmon FP. A Tunable Measure for Information Leakage. In 2018 IEEE International Symposium on Information Theory, ISIT 2018. Vol. 2018-June. Institute of Electrical and Electronics Engineers Inc. 2018. p. 701-705. 8437307 https://doi.org/10.1109/ISIT.2018.8437307
Liao, Jiachun ; Kosut, Oliver ; Sankar, Lalitha ; Calmon, Flavio P. / A Tunable Measure for Information Leakage. 2018 IEEE International Symposium on Information Theory, ISIT 2018. Vol. 2018-June Institute of Electrical and Electronics Engineers Inc., 2018. pp. 701-705
@inproceedings{fffc635985d34020adb52b4652ddb0a0,
title = "A Tunable Measure for Information Leakage",
abstract = "A tunable measure for information leakage called maximal a-leakage is introduced. This measure quantifies the maximal gain of an adversary in refining a tilted version of its prior belief of any (potentially random) function of a dataset conditioning on a disclosed dataset. The choice of \alpha determines the specific adversarial action ranging from refining a belief for \alpha=1 to guessing the best posterior for \alpha=\infty, and for these extremal values this measure simplifies to mutual information (MI) and maximal leakage (MaxL), respectively. For all other \alpha this measure is shown to be the Arimoto channel capacity. Several properties of this measure are proven including: (i) quasi-convexity in the mapping between the original and disclosed datasets; (ii) data processing inequalities; and (iii) a composition property. A full version of this paper is in [1].",
author = "Jiachun Liao and Oliver Kosut and Lalitha Sankar and Calmon, {Flavio P.}",
year = "2018",
month = "8",
day = "15",
doi = "10.1109/ISIT.2018.8437307",
language = "English (US)",
isbn = "9781538647806",
volume = "2018-June",
pages = "701--705",
booktitle = "2018 IEEE International Symposium on Information Theory, ISIT 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - A Tunable Measure for Information Leakage

AU - Liao, Jiachun

AU - Kosut, Oliver

AU - Sankar, Lalitha

AU - Calmon, Flavio P.

PY - 2018/8/15

Y1 - 2018/8/15

N2 - A tunable measure for information leakage called maximal a-leakage is introduced. This measure quantifies the maximal gain of an adversary in refining a tilted version of its prior belief of any (potentially random) function of a dataset conditioning on a disclosed dataset. The choice of \alpha determines the specific adversarial action ranging from refining a belief for \alpha=1 to guessing the best posterior for \alpha=\infty, and for these extremal values this measure simplifies to mutual information (MI) and maximal leakage (MaxL), respectively. For all other \alpha this measure is shown to be the Arimoto channel capacity. Several properties of this measure are proven including: (i) quasi-convexity in the mapping between the original and disclosed datasets; (ii) data processing inequalities; and (iii) a composition property. A full version of this paper is in [1].

AB - A tunable measure for information leakage called maximal a-leakage is introduced. This measure quantifies the maximal gain of an adversary in refining a tilted version of its prior belief of any (potentially random) function of a dataset conditioning on a disclosed dataset. The choice of \alpha determines the specific adversarial action ranging from refining a belief for \alpha=1 to guessing the best posterior for \alpha=\infty, and for these extremal values this measure simplifies to mutual information (MI) and maximal leakage (MaxL), respectively. For all other \alpha this measure is shown to be the Arimoto channel capacity. Several properties of this measure are proven including: (i) quasi-convexity in the mapping between the original and disclosed datasets; (ii) data processing inequalities; and (iii) a composition property. A full version of this paper is in [1].

UR - http://www.scopus.com/inward/record.url?scp=85052481906&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052481906&partnerID=8YFLogxK

U2 - 10.1109/ISIT.2018.8437307

DO - 10.1109/ISIT.2018.8437307

M3 - Conference contribution

SN - 9781538647806

VL - 2018-June

SP - 701

EP - 705

BT - 2018 IEEE International Symposium on Information Theory, ISIT 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -