#### DMCA

## Differentially private histogram publication (2012)

Venue: | In ICDE |

Citations: | 26 - 3 self |

### Citations

647 | Differential privacy - Dwork - 2006 |

646 | A.: Calibrating noise to sensitivity in private data analysis
- Dwork, McSherry, et al.
- 2006
(Show Context)
Citation Context ...reatens the publication of any research paper on genome-wide association studies, which currently is an active field in biomedical research. The recently proposed concept of differential privacy (DP) =-=[3]-=-–[8] addresses the above issues by injecting a small amount of random noise into statistical results. DP is rapidly gaining popularity, because it provides rigorous privacy guarantees against adversar... |

256 | Differential privacy: a survey of results
- Dwork
- 2008
(Show Context)
Citation Context ....38 12,494 124 1.51 StructureFirst 566,212 3715 36 839,892 9,051 88 1,240,337 12,192 115 1,970,914 12,430 119 TABLE I COMPARISON OF THE AVERAGE SQUARE ERRORS ON QUERY WITH UNIT LENGTH, I.E. ( SSE n ) =-=[13]-=- for surveys). For example, Bhaskar et al. [14] investigate how frequent itemsets from transaction data can be published. Friedman et al. [15] devise methods for constructing decision trees. Korolova ... |

218 | A learning theory approach to non-interactive database privacy
- Blum, Ligett, et al.
- 2008
(Show Context)
Citation Context ...or publishing statistics in search logs, while McSherry and Mahajan [18] develop techniques for network trace analysis. Among the existing approaches, the ones most related to ours are by Blum et al. =-=[19]-=-, Hay et al. [6], Xiao et al. [8], and Li et al. [7]. Specifically, Blum et al. [19] propose to construct one-dimensional histograms by dividing the input counts into several bins, such that the sum o... |

171 | Optimal histograms with quality guarantees
- JAGADISH, POOSALA, et al.
- 1998
(Show Context)
Citation Context ...70 i=1 2 3 4 5 6 7 k=1 2 3 Fig. 3. Build of an optimal histogram on Figure 1(b). Each entry T (i, k) keeps the minimal error for first i counts with exactly k bins conventional Histogram Construction =-=[10]-=-, [11] is to identify the optimal histogram structure to minimize Error(H,D), as formalized in the following. Problem 1. Given the count sequence D and the size constraint k, find the optimal histogra... |

108 |
Resolving individuals contributing trace amounts of DNA to highly complex mixtures using high-density SNP genotyping microarrays.
- Homer, Szelinger, et al.
- 2008
(Show Context)
Citation Context ...s complicate the publication of analysis results. A notable example is the dbGaP1 database, which contains results of genetic studies. Such results used to be publicly available, until a recent paper =-=[1]-=- describes an attack that infers whether a person has participated in a certain study (e.g., on patients with 1http://www.ncbi.nlm.nih.gov/gap Name Age HIV+ Alice 42 Yes Bob 31 Yes Carol 32 Yes Dave 3... |

104 | The Laplace Distribution and Generalizations: A Revisit with Applications to - Kotz, Kozubowski, et al. - 2001 |

100 | Differential privacy via wavelet transforms
- Xiao, Wang, et al.
- 2010
(Show Context)
Citation Context ...ens the publication of any research paper on genome-wide association studies, which currently is an active field in biomedical research. The recently proposed concept of differential privacy (DP) [3]–=-=[8]-=- addresses the above issues by injecting a small amount of random noise into statistical results. DP is rapidly gaining popularity, because it provides rigorous privacy guarantees against adversaries ... |

96 | Optimizing linear counting queries under differential privacy
- Li, Hay, et al.
- 2010
(Show Context)
Citation Context ...ry and Mahajan [18] develop techniques for network trace analysis. Among the existing approaches, the ones most related to ours are by Blum et al. [19], Hay et al. [6], Xiao et al. [8], and Li et al. =-=[7]-=-. Specifically, Blum et al. [19] propose to construct one-dimensional histograms by dividing the input counts into several bins, such that the sum of counts in each bin is roughly the same. The bin co... |

88 | Differentially Private Aggregation of Distributed Time-Series with Transformation and Encryption
- Rastogi, Nath
(Show Context)
Citation Context ...(more) noise, which leads to reduced relative errors. This approach, however, is inapplicable for our problem, since the count queries concerned in a histogram are mutually disjoint. Rastogi and Nath =-=[23]-=- develop a technique for releasing aggregated results on time series data collected from distributed users. The technique injects noise into a time series by first (i) deriving an approximation of the... |

72 | The price of privacy and the limits of lp decoding - DWORK, MCSHERRY, et al. |

67 | Boosting the accuracy of differentially private histograms through consistency.
- Hay, Rastogi, et al.
- 2010
(Show Context)
Citation Context ...y of an already published histogram using an existing method [3]. Moreover, we adapt DP-histograms to answer arbitrary range-count queries, which has drawn considerable research attention (e.g., [3], =-=[6]-=-). For such queries, NoiseFirst achieves better accuracy for short ranges, whereas StructureFirst is more suitable for longer ones. Extensive experiments using several real data sets demonstrate that ... |

64 |
Releasing search queries and clicks privately.
- Korolova, Kenthapadi, et al.
- 2009
(Show Context)
Citation Context ...rveys). For example, Bhaskar et al. [14] investigate how frequent itemsets from transaction data can be published. Friedman et al. [15] devise methods for constructing decision trees. Korolova et al. =-=[16]-=- and Götz et al. [17] present methods for publishing statistics in search logs, while McSherry and Mahajan [18] develop techniques for network trace analysis. Among the existing approaches, the ones ... |

46 | Private and continual release of statistics
- Chan, Shi, et al.
- 2010
(Show Context)
Citation Context ...dard solution, by merging adjacent noisy counts. Intuitively, averaging over the neighboring noisy counts is able to eliminate the impact of zero-mean Laplace noise, based on the large number theorem =-=[12]-=-. In the rest of the section, we provide some theoretical analysis on the expected error incurred by NoiseFirst. The major challenge in the analysis is deriving the connection between (i) the error of... |

43 | Data mining with differential privacy
- Friedman, Schuster
(Show Context)
Citation Context ...E SQUARE ERRORS ON QUERY WITH UNIT LENGTH, I.E. ( SSE n ) [13] for surveys). For example, Bhaskar et al. [14] investigate how frequent itemsets from transaction data can be published. Friedman et al. =-=[15]-=- devise methods for constructing decision trees. Korolova et al. [16] and Götz et al. [17] present methods for publishing statistics in search logs, while McSherry and Mahajan [18] develop techniques... |

41 | Approximation and streaming algorithms for histogram construction problems.
- Guha, Koudas, et al.
- 2006
(Show Context)
Citation Context ... 2 3 4 5 6 7 k=1 2 3 Fig. 3. Build of an optimal histogram on Figure 1(b). Each entry T (i, k) keeps the minimal error for first i counts with exactly k bins conventional Histogram Construction [10], =-=[11]-=- is to identify the optimal histogram structure to minimize Error(H,D), as formalized in the following. Problem 1. Given the count sequence D and the size constraint k, find the optimal histogram H∗ t... |

41 | Differentially-private network trace analysis
- MCSHERRY, MAHAJAN
- 2010
(Show Context)
Citation Context ...d. Friedman et al. [15] devise methods for constructing decision trees. Korolova et al. [16] and Götz et al. [17] present methods for publishing statistics in search logs, while McSherry and Mahajan =-=[18]-=- develop techniques for network trace analysis. Among the existing approaches, the ones most related to ours are by Blum et al. [19], Hay et al. [6], Xiao et al. [8], and Li et al. [7]. Specifically, ... |

40 | Differentially private data cubes: optimizing noise sources and consistency.
- Ding, Winslett, et al.
- 2011
(Show Context)
Citation Context ...mputation cost, and hence is inapplicable on large data sets. In addition, there exist several techniques that address problems similar to (but different from) ours. Barak et al. [20] and Ding et al. =-=[21]-=- propose methods for releasing marginals, i.e., projections of a data set onto subsets of its attributes. The core ideas of their methods are to exploit the correlations among the marginals to reduce ... |

37 | Discovering frequent patterns in sensitive data. In
- Bhaskar, Laxman, et al.
- 2010
(Show Context)
Citation Context ... 36 839,892 9,051 88 1,240,337 12,192 115 1,970,914 12,430 119 TABLE I COMPARISON OF THE AVERAGE SQUARE ERRORS ON QUERY WITH UNIT LENGTH, I.E. ( SSE n ) [13] for surveys). For example, Bhaskar et al. =-=[14]-=- investigate how frequent itemsets from transaction data can be published. Friedman et al. [15] devise methods for constructing decision trees. Korolova et al. [16] and Götz et al. [17] present metho... |

35 | Learning your identity and disease from research papers: Information leaks in genome wide association study.
- Wang, Li, et al.
- 2009
(Show Context)
Citation Context ... 1. Example sensitive dataset and its corresponding histogram diabetes) from its results; thereafter, access to such results is strictly controlled. Furthermore, a strengthened version of this attack =-=[2]-=- threatens the publication of any research paper on genome-wide association studies, which currently is an active field in biomedical research. The recently proposed concept of differential privacy (D... |

32 | Ireduct: differential privacy with reduced relative errors
- Xiao, Bender, et al.
- 2011
(Show Context)
Citation Context ...for privacy protection. However, neither Barak et al.’s nor Ding et al.’s method can be applied for our problem, as we consider the release of one histogram instead of multiple marginals. Xiao et al. =-=[22]-=- devise a differentially private approach that optimizes the relative errors of a given set of count queries. The approach targets the scenario where the count queries overlap with each other (i.e., t... |

23 |
consistency too: a holistic solution to contingency table release
- BARAK, CHAUDHURI, et al.
(Show Context)
Citation Context ...incurs significant computation cost, and hence is inapplicable on large data sets. In addition, there exist several techniques that address problems similar to (but different from) ours. Barak et al. =-=[20]-=- and Ding et al. [21] propose methods for releasing marginals, i.e., projections of a data set onto subsets of its attributes. The core ideas of their methods are to exploit the correlations among the... |

17 | Publishing search logs - a comparative study of privacy guarantees.
- Götz, Machanavajjhala, et al.
- 2011
(Show Context)
Citation Context ...haskar et al. [14] investigate how frequent itemsets from transaction data can be published. Friedman et al. [15] devise methods for constructing decision trees. Korolova et al. [16] and Götz et al. =-=[17]-=- present methods for publishing statistics in search logs, while McSherry and Mahajan [18] develop techniques for network trace analysis. Among the existing approaches, the ones most related to ours a... |