Max-Information, Differential Privacy, and Post-selection Hypothesis Testing

@article{Rogers2016MaxInformationDP,
  title={Max-Information, Differential Privacy, and Post-selection Hypothesis Testing},
  author={R. Rogers and A. Roth and A. D. Smith and Om Thakkar},
  journal={2016 IEEE 57th Annual Symposium on Foundations of Computer Science (FOCS)},
  year={2016},
  pages={487-494}
}
  • R. Rogers, A. Roth, +1 author Om Thakkar
  • Published 2016
  • Computer Science, Mathematics
  • 2016 IEEE 57th Annual Symposium on Foundations of Computer Science (FOCS)
  • In this paper, we initiate a principled study of how the generalization properties of approximate differential privacy can be used to perform adaptive hypothesis testing, while giving statistically valid p-value corrections. We do this by observing that the guarantees of algorithms with bounded approximate max-information are sufficient to correct the p-values of adaptively chosen hypotheses, and then by proving that algorithms that satisfy (∈,δ)-differential privacy have bounded approximate… CONTINUE READING

    Topics from this paper.

    Mitigating Bias in Adaptive Data Gathering via Differential Privacy
    • 11
    • Open Access
    Differential Privacy for Sequential Algorithms
    Leveraging Privacy In Data Analysis
    • 3
    Lower Bounds for Locally Private Estimation via Communication Complexity
    • 31
    • Open Access
    Generalization Error Bounds Via Rényi-, f-Divergences and Maximal Leakage
    • 4
    • Highly Influenced
    • Open Access
    Heavy Hitters and the Structure of Local Privacy
    • 71
    • Open Access
    Heavy Hitters and the Structure of Local Privacy
    • 1
    • Open Access
    Learners that Leak Little Information
    • 6
    • Open Access

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 47 REFERENCES
    Differentially Private Hypothesis Testing, Revisited
    • 43
    • Open Access
    The Limits of Two-Party Differential Privacy
    • 153
    • Open Access
    The Algorithmic Foundations of Differential Privacy
    • 2,211
    • Open Access
    Inference using noisy degrees: Differentially private $\beta$-model and synthetic graphs
    • 62
    • Open Access
    Differentially Private Chi-Squared Hypothesis Testing: Goodness of Fit and Independence Testing
    • 65
    Lower Bounds in Differential Privacy
    • A. De
    • Computer Science
    • 2012
    • 93
    • Highly Influential
    • Open Access
    Interactive privacy via the median mechanism
    • 242
    • Open Access
    Private False Discovery Rate Control
    • 19
    • Open Access