The VLDB Journal

, Volume 24, Issue 6, pp 757–781 | Cite as

The matrix mechanism: optimizing linear counting queries under differential privacy

  • Chao Li
  • Gerome Miklau
  • Michael Hay
  • Andrew McGregor
  • Vibhor Rastogi
Regular Paper


Differential privacy is a robust privacy standard that has been successfully applied to a range of data analysis tasks. We describe the matrix mechanism, an algorithm for answering a workload of linear counting queries that adapts the noise distribution to properties of the provided queries. Given a workload, the mechanism uses a different set of queries, called a query strategy, which are answered using a standard Laplace or Gaussian mechanism. Noisy answers to the workload queries are then derived from the noisy answers to the strategy queries. This two-stage process can result in a more complex, correlated noise distribution that preserves differential privacy but increases accuracy. We provide a formal analysis of the error of query answers produced by the mechanism and investigate the problem of computing the optimal query strategy in support of a given workload. We show that this problem can be formulated as a rank-constrained semidefinite program. We analyze two seemingly distinct techniques proposed in the literature, whose similar behavior is explained by viewing them as instances of the matrix mechanism. We also describe an extension of the mechanism in which nonnegativity constraints are included in the derivation process and provide experimental evidence of its efficacy.


Differential privacy Linear query Matrix mechanism Semidefinite program Least squares 


  1. 1.
    Ács, G., Castelluccia, C., Chen, R.: Differentially private histogram publishing through lossy compression. In: ICDM, pp. 1–10 (2012)Google Scholar
  2. 2.
    Barak, B., Chaudhuri, K., Dwork, C., Kale, S., McSherry, F., Talwar, K.: Privacy, accuracy, and consistency too: a holistic solution to contingency table release. In: PODS (2007)Google Scholar
  3. 3.
    Ben-Israel, A., Greville, T.: Generalized Inverses: Theory and Applications, vol. 15. Springer, Berlin (2003)Google Scholar
  4. 4.
    Cormode, G., Procopiuc, M., Shen, E., Srivastava, D., Yu, T.: Differentially private spatial decompositions. In: ICDE (2012)Google Scholar
  5. 5.
    Dattorro, J.: Convex Optimization & Euclidean Distance Geometry. Meboo Publishing, USA (2005)Google Scholar
  6. 6.
    Ding, B., Winslett, M., Han, J., Li, Z.: Differentially private data cubes: optimizing noise sources and consistency. In: SIGMOD, pp. 217–228 (2011)Google Scholar
  7. 7.
    Dwork, C.: Differential privacy: a survey of results. In: TAMC (2008)Google Scholar
  8. 8.
    Dwork, C.: The differential privacy frontier. In: TCC (2009)Google Scholar
  9. 9.
    Dwork, C.: A firm foundation for private data analysis. Commun. ACM 54(1), 86–95 (2011)CrossRefGoogle Scholar
  10. 10.
    Dwork, C., Kenthapadi, K., McSherry, F., Mironov, I., Naor, M.: Our data, ourselves: privacy via distributed noise generation. In: EUROCRYPT, pp. 486–503 (2006)Google Scholar
  11. 11.
    Dwork, C., Naor, M., Reingold, O., Rothblum, G., Vadhan, S.: On the complexity of differentially private data release: efficient algorithms and hardness results. In: STOC, pp. 381–390 (2009)Google Scholar
  12. 12.
    Dwork, C., Nissim, F.M.K., Smith, A.: Calibrating noise to sensitivity in private data analysis. In: TCC (2006)Google Scholar
  13. 13.
    Dwork, C., Rothblum, G.N., Vadhan, S.P.: Boosting and differential privacy. In: FOCS, pp. 51–60 (2010)Google Scholar
  14. 14.
    Ghosh, A., Roughgarden, T., Sundararajan, M.: Universally utility-maximizing privacy mechanisms. In: STOC (2009)Google Scholar
  15. 15.
    Gupta, A., Roth, A., Ullman, J.: Iterative constructions and private data release. In: TCC, pp. 339–356 (2012)Google Scholar
  16. 16.
    Hardt, M., Ligett, K., McSherry, F.: A simple and practical algorithm for differentially private data release. In: NIPS, pp. 2348–2356 (2012)Google Scholar
  17. 17.
    Hardt, M., Rothblum, G.: A multiplicative weights mechanism for privacy-preserving data analysis. In: FOCS, pp. 61–70 (2010)Google Scholar
  18. 18.
    Hardt, M., Talwar, K.: On the geometry of differential privacy. In: STOC, pp. 705–714 (2010)Google Scholar
  19. 19.
    Hay, M., Rastogi, V., Miklau, G., Suciu, D.: Boosting the accuracy of differentially-private histograms through consistency. PVLDB 3(1–2), 1021–1032 (2010)Google Scholar
  20. 20.
    Li, C., Hay, M., Rastogi, V., Miklau, G., McGregor, A.: Optimizing linear counting queries under differential privacy. In: PODS, pp. 123–134 (2010)Google Scholar
  21. 21.
    Li, C., Miklau, G.: An adaptive mechanism for accurate query answering under differential privacy. PVLDB 5(6), 514–525 (2012)Google Scholar
  22. 22.
    McSherry, F., Mironov, I.: Differentially private recommender systems: building privacy into the netflix prize contenders. In: SIGKDD (2009)Google Scholar
  23. 23.
    McSherry, F.D.: Privacy integrated queries: an extensible platform for privacy-preserving data analysis. In: SIGMOD, pp. 19–30 (2009)Google Scholar
  24. 24.
    Nikolov, A., Talwar, K., Zhang, L.: The geometry of differential privacy: the sparse and approximate cases. In: STOC (2013)Google Scholar
  25. 25.
    Nissim, K., Raskhodnikova, S., Smith, A.: Smooth sensitivity and sampling in private data analysis. In: STOC, pp. 75–84 (2007)Google Scholar
  26. 26.
    Qardaji, W.H., Yang, W., Li, N.: Understanding hierarchical methods for differentially private histograms. PVLDB 6(14), 1954–1965 (2013)Google Scholar
  27. 27.
    Roth, A., Roughgarden, T.: Interactive privacy via the median mechanism. In: STOC, pp. 765–774 (2010)Google Scholar
  28. 28.
    Xiao, X., Wang, G., Gehrke, J.: Differential privacy via wavelet transforms. In: ICDE, pp. 225–236 (2010)Google Scholar
  29. 29.
    Xiao, Y., Gardner, J.J., Xiong, L.: Dpcube: Releasing differentially private data cubes for health information. In: ICDE, pp. 1305–1308 (2012)Google Scholar
  30. 30.
    Xu, J., Zhang, Z., Xiao, X., Yang, Y., Yu, G., Winslett, M.: Differentially private histogram publication. VLDB J 22(6), 797–822 (2013)CrossRefGoogle Scholar
  31. 31.
    Yaroslavtsev, G., Cormode, G., Procopiuc, C.M., Srivastava, D.: Accurate and efficient private release of datacubes and contingency tables. In: ICDE (2013)Google Scholar
  32. 32.
    Yuan, G., Zhang, Z., Winslett, M., Xiao, X., Yang, Y., Hao, Z.: Low-rank mechanism: optimizing batch queries under differential privacy. In: VLDB (2012)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2015

Authors and Affiliations

  1. 1.College of Information and Computer SciencesUniversity of Massachusetts AmherstAmherstUSA
  2. 2.Department of Computer ScienceColgate UniversityHamiltonUSA
  3. 3.Twitter Inc.San FranciscoUSA
  4. 4.Google Inc.Mountain ViewUSA

Personalised recommendations