Analyzing graphs with node differential privacy

Shiva Prasad Kasiviswanathan, Kobbi Nissim, Sofya Raskhodnikova, Adam Davison Smith

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

65 Citations (Scopus)

Abstract

We develop algorithms for the private analysis of network data that provide accurate analysis of realistic networks while satisfying stronger privacy guarantees than those of previous work. We present several techniques for designing node differentially private algorithms, that is, algorithms whose output distribution does not change significantly when a node and all its adjacent edges are added to a graph. We also develop methodology for analyzing the accuracy of such algorithms on realistic networks. The main idea behind our techniques is to "project" (in one of several senses) the input graph onto the set of graphs with maximum degree below a certain threshold. We design projection operators, tailored to specific statistics that have low sensitivity and preserve information about the original statistic. These operators can be viewed as giving a fractional (low-degree) graph that is a solution to an optimization problem described as a maximum flow instance, linear program, or convex program. In addition, we derive a generic, efficient reduction that allows us to apply any differentially private algorithm for bounded-degree graphs to an arbitrary graph. This reduction is based on analyzing the smooth sensitivity of the "naive" truncation that simply discards nodes of high degree.

Original languageEnglish (US)
Title of host publicationTheory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings
Pages457-476
Number of pages20
DOIs
StatePublished - Feb 21 2013
Event10th Theory of Cryptography Conference, TCC 2013 - Tokyo, Japan
Duration: Mar 3 2013Mar 6 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7785 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other10th Theory of Cryptography Conference, TCC 2013
CountryJapan
CityTokyo
Period3/3/133/6/13

Fingerprint

Privacy
Graph in graph theory
Vertex of a graph
Mathematical operators
Statistics
Convex Program
Maximum Flow
Projection Operator
Maximum Degree
Linear Program
Truncation
Statistic
Fractional
Adjacent
Optimization Problem
Methodology
Output
Arbitrary
Operator

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Kasiviswanathan, S. P., Nissim, K., Raskhodnikova, S., & Smith, A. D. (2013). Analyzing graphs with node differential privacy. In Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings (pp. 457-476). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7785 LNCS). https://doi.org/10.1007/978-3-642-36594-2_26
Kasiviswanathan, Shiva Prasad ; Nissim, Kobbi ; Raskhodnikova, Sofya ; Smith, Adam Davison. / Analyzing graphs with node differential privacy. Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings. 2013. pp. 457-476 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{154060bceeda4890b8bb16439d2639f6,
title = "Analyzing graphs with node differential privacy",
abstract = "We develop algorithms for the private analysis of network data that provide accurate analysis of realistic networks while satisfying stronger privacy guarantees than those of previous work. We present several techniques for designing node differentially private algorithms, that is, algorithms whose output distribution does not change significantly when a node and all its adjacent edges are added to a graph. We also develop methodology for analyzing the accuracy of such algorithms on realistic networks. The main idea behind our techniques is to {"}project{"} (in one of several senses) the input graph onto the set of graphs with maximum degree below a certain threshold. We design projection operators, tailored to specific statistics that have low sensitivity and preserve information about the original statistic. These operators can be viewed as giving a fractional (low-degree) graph that is a solution to an optimization problem described as a maximum flow instance, linear program, or convex program. In addition, we derive a generic, efficient reduction that allows us to apply any differentially private algorithm for bounded-degree graphs to an arbitrary graph. This reduction is based on analyzing the smooth sensitivity of the {"}naive{"} truncation that simply discards nodes of high degree.",
author = "Kasiviswanathan, {Shiva Prasad} and Kobbi Nissim and Sofya Raskhodnikova and Smith, {Adam Davison}",
year = "2013",
month = "2",
day = "21",
doi = "10.1007/978-3-642-36594-2_26",
language = "English (US)",
isbn = "9783642365935",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "457--476",
booktitle = "Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings",

}

Kasiviswanathan, SP, Nissim, K, Raskhodnikova, S & Smith, AD 2013, Analyzing graphs with node differential privacy. in Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7785 LNCS, pp. 457-476, 10th Theory of Cryptography Conference, TCC 2013, Tokyo, Japan, 3/3/13. https://doi.org/10.1007/978-3-642-36594-2_26

Analyzing graphs with node differential privacy. / Kasiviswanathan, Shiva Prasad; Nissim, Kobbi; Raskhodnikova, Sofya; Smith, Adam Davison.

Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings. 2013. p. 457-476 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7785 LNCS).

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

TY - GEN

T1 - Analyzing graphs with node differential privacy

AU - Kasiviswanathan, Shiva Prasad

AU - Nissim, Kobbi

AU - Raskhodnikova, Sofya

AU - Smith, Adam Davison

PY - 2013/2/21

Y1 - 2013/2/21

N2 - We develop algorithms for the private analysis of network data that provide accurate analysis of realistic networks while satisfying stronger privacy guarantees than those of previous work. We present several techniques for designing node differentially private algorithms, that is, algorithms whose output distribution does not change significantly when a node and all its adjacent edges are added to a graph. We also develop methodology for analyzing the accuracy of such algorithms on realistic networks. The main idea behind our techniques is to "project" (in one of several senses) the input graph onto the set of graphs with maximum degree below a certain threshold. We design projection operators, tailored to specific statistics that have low sensitivity and preserve information about the original statistic. These operators can be viewed as giving a fractional (low-degree) graph that is a solution to an optimization problem described as a maximum flow instance, linear program, or convex program. In addition, we derive a generic, efficient reduction that allows us to apply any differentially private algorithm for bounded-degree graphs to an arbitrary graph. This reduction is based on analyzing the smooth sensitivity of the "naive" truncation that simply discards nodes of high degree.

AB - We develop algorithms for the private analysis of network data that provide accurate analysis of realistic networks while satisfying stronger privacy guarantees than those of previous work. We present several techniques for designing node differentially private algorithms, that is, algorithms whose output distribution does not change significantly when a node and all its adjacent edges are added to a graph. We also develop methodology for analyzing the accuracy of such algorithms on realistic networks. The main idea behind our techniques is to "project" (in one of several senses) the input graph onto the set of graphs with maximum degree below a certain threshold. We design projection operators, tailored to specific statistics that have low sensitivity and preserve information about the original statistic. These operators can be viewed as giving a fractional (low-degree) graph that is a solution to an optimization problem described as a maximum flow instance, linear program, or convex program. In addition, we derive a generic, efficient reduction that allows us to apply any differentially private algorithm for bounded-degree graphs to an arbitrary graph. This reduction is based on analyzing the smooth sensitivity of the "naive" truncation that simply discards nodes of high degree.

UR - http://www.scopus.com/inward/record.url?scp=84873947082&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84873947082&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-36594-2_26

DO - 10.1007/978-3-642-36594-2_26

M3 - Conference contribution

SN - 9783642365935

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

SP - 457

EP - 476

BT - Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings

ER -

Kasiviswanathan SP, Nissim K, Raskhodnikova S, Smith AD. Analyzing graphs with node differential privacy. In Theory of Cryptography - 10th Theory of Cryptography Conference, TCC 2013, Proceedings. 2013. p. 457-476. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-36594-2_26