Asymptotically optimal and private statistical estimation

Adam Smith

doi:10.1007/978-3-642-10433-6_4

Asymptotically optimal and private statistical estimation

Adam Smith

School of Electrical Engineering and Computer Science

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

6 Scopus citations

Abstract

Differential privacy is a definition of "privacy" for statistical databases. The definition is simple, yet it implies strong semantics even in the presence of an adversary with arbitrary auxiliary information about the database. In this talk, we discuss recent work on measuring the utility of differentially private analyses via the traditional yardsticks of statistical inference. Specifically, we discuss two differentially private estimators that, given i.i.d. samples from a probability distribution, converge to the correct answer at the same rate as the optimal nonprivate estimator.

Original language	English (US)
Title of host publication	Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings
Pages	53-57
Number of pages	5
DOIs	https://doi.org/10.1007/978-3-642-10433-6_4
State	Published - 2009
Event	8th International Conference on Cryptology and Network Security, CANS 2009 - Kanazawa, Japan Duration: Dec 12 2009 → Dec 14 2009

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	5888 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	8th International Conference on Cryptology and Network Security, CANS 2009
Country/Territory	Japan
City	Kanazawa
Period	12/12/09 → 12/14/09

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-10433-6_4

Cite this

Smith, A. (2009). Asymptotically optimal and private statistical estimation. In Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings (pp. 53-57). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5888 LNCS). https://doi.org/10.1007/978-3-642-10433-6_4

@inproceedings{848bd53fb68d46f8845dbb3093b94207,

title = "Asymptotically optimal and private statistical estimation",

abstract = "Differential privacy is a definition of {"}privacy{"} for statistical databases. The definition is simple, yet it implies strong semantics even in the presence of an adversary with arbitrary auxiliary information about the database. In this talk, we discuss recent work on measuring the utility of differentially private analyses via the traditional yardsticks of statistical inference. Specifically, we discuss two differentially private estimators that, given i.i.d. samples from a probability distribution, converge to the correct answer at the same rate as the optimal nonprivate estimator.",

author = "Adam Smith",

note = "Funding Information: Supported in part by US National Science Foundation (NSF) TF award #0747294 and NSF CAREER/PECASE award #0729171.; 8th International Conference on Cryptology and Network Security, CANS 2009 ; Conference date: 12-12-2009 Through 14-12-2009",

year = "2009",

doi = "10.1007/978-3-642-10433-6_4",

language = "English (US)",

isbn = "3642104320",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "53--57",

booktitle = "Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings",

}

Smith, A 2009, Asymptotically optimal and private statistical estimation. in Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5888 LNCS, pp. 53-57, 8th International Conference on Cryptology and Network Security, CANS 2009, Kanazawa, Japan, 12/12/09. https://doi.org/10.1007/978-3-642-10433-6_4

Asymptotically optimal and private statistical estimation. / Smith, Adam.
Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings. 2009. p. 53-57 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5888 LNCS).

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

TY - GEN

T1 - Asymptotically optimal and private statistical estimation

AU - Smith, Adam

N1 - Funding Information: Supported in part by US National Science Foundation (NSF) TF award #0747294 and NSF CAREER/PECASE award #0729171.

PY - 2009

Y1 - 2009

N2 - Differential privacy is a definition of "privacy" for statistical databases. The definition is simple, yet it implies strong semantics even in the presence of an adversary with arbitrary auxiliary information about the database. In this talk, we discuss recent work on measuring the utility of differentially private analyses via the traditional yardsticks of statistical inference. Specifically, we discuss two differentially private estimators that, given i.i.d. samples from a probability distribution, converge to the correct answer at the same rate as the optimal nonprivate estimator.

AB - Differential privacy is a definition of "privacy" for statistical databases. The definition is simple, yet it implies strong semantics even in the presence of an adversary with arbitrary auxiliary information about the database. In this talk, we discuss recent work on measuring the utility of differentially private analyses via the traditional yardsticks of statistical inference. Specifically, we discuss two differentially private estimators that, given i.i.d. samples from a probability distribution, converge to the correct answer at the same rate as the optimal nonprivate estimator.

UR - http://www.scopus.com/inward/record.url?scp=71549153186&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=71549153186&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-10433-6_4

DO - 10.1007/978-3-642-10433-6_4

M3 - Conference contribution

AN - SCOPUS:71549153186

SN - 3642104320

SN - 9783642104329

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 53

EP - 57

BT - Cryptology and Network Security - 8th International Conference, CANS 2009, Proceedings

T2 - 8th International Conference on Cryptology and Network Security, CANS 2009

Y2 - 12 December 2009 through 14 December 2009

ER -

Asymptotically optimal and private statistical estimation

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this