Pith. sign in

REVIEW 4 major objections 4 minor 19 references

A Novel Approach to Differential Privacy with Alpha Divergence

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Alpha-divergence privacy claims tighter budgets for short iterative analyses.

desk verdict This is a monotone reparameterization of Rényi differential privacy, not a new framework; the headline advantage is an artifact and the composition proof has a real error. read the letter →

arxiv 2506.17012 v1 pith:FM53EZBS submitted 2025-06-20 cs.CR

Yifeng Liu , Zehua Wang This is my paper · ORCID
classification cs.CR MSC 68P2794A17
keywords alphadivergencedifferentialprivacyRényizero-concentratedGaussianmechanismcompositionaccountingadaptive
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Alpha differential privacy (ADP) defines a mechanism's privacy loss as the alpha divergence between its output distributions on adjacent datasets, requiring that divergence to stay below epsilon. The paper derives ADP's basic properties—post-processing, adaptive composition, group privacy—and gives closed-form ADP parameters for randomized response, Laplace, and Gaussian mechanisms. The paper's central claim is that this accounting starts from a lower per-query privacy cost than Rényi differential privacy (RDP), zero-concentrated differential privacy (zCDP), and advanced composition, so cumulative privacy consumption is smaller in settings with few iterations and very strict failure probabilities. Simulations with the Gaussian mechanism are presented as evidence, with recommended alpha selection rules for practitioners. The authors also acknowledge that the advantage fades at very large iteration counts, where RDP and zCDP become more favorable.

What carries the argument

The central object is $\alpha$ divergence, the $f$-divergence generated by $f(u) = \frac{u^\alpha - \alpha u - (1-\alpha)}{\alpha(\alpha-1)}$, written as $eD_\alpha(P\|Q) = \frac{1}{\alpha(\alpha-1)}(\int p^\alpha q^{1-\alpha}\,d\lambda - 1)$. This quantity is what ADP bounds by $\epsilon$, and it drives the composition formula: composing two $(\alpha,\epsilon)$-ADP steps adds costs with a cross term $\alpha(\alpha-1)\epsilon_1\epsilon_2$. The machinery also includes the conversion rule to $(\epsilon,\delta)$-DP and the per-mechanism closed forms, which together let the paper's $\alpha$-selection algorithms search over $\alpha$ to minimize cumulative privacy consumption.

What would settle it

Take the paper's Gaussian ADP bound, plug it into the identity $D_\alpha^{\text{RDP}} = \frac{1}{\alpha-1}\log(1+\alpha(\alpha-1)eD_\alpha)$, and check whether it equals $\alpha\Delta^2/(2\sigma^2)$; if it does, ADP's per-query Gaussian guarantee is identical to RDP's, so any reported advantage in cumulative curves is a consequence of parameter choice rather than a new privacy measure.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that if a mechanism $M$ satisfies $eD_\alpha(M(D)\|M(D')) \le \epsilon$ for all adjacent $D,D'$, then it is $(\alpha,\epsilon)$-ADP, and this condition supports a set of privacy-preserving properties. The adaptive composition rule states that two mechanisms with parameters $(\alpha,\epsilon_1)$ and $(\alpha,\epsilon_2)$ compose to $(\alpha, \epsilon_1+\epsilon_2+\alpha(\alpha-1)\epsilon_1\epsilon_2)$-ADP. For the Gaussian mechanism with $\ell_2$ sensitivity $\Delta$ and variance $\sigma^2$, the paper derives the per-query cost $\epsilon = \frac{1}{\alpha(\alpha-1)}(\exp(\alpha(\alpha-1)\Delta^2/(2\sigma^2))-1)$, and it converts ADP to $(\epsilon,\delta)$-DP via a Markov-style argument. The paper claims that this framework yields lower cumulative privacy consumption than RDP, zCDP, and advanced composition in small-to-moderate iterative settings, especially under strict failure probabilities, while recognizing that RDP and zCDP may be better for very many iterations.

Load-bearing premise

The framework's claimed advantage depends on treating alpha-divergence privacy as a genuinely different accounting method from Rényi differential privacy, even though the two are linked by a direct identity that converts the paper's Gaussian ADP bound into the standard RDP Gaussian bound.

Editorial extensions

If this is right

  • ADP gives a per-query privacy cost for Gaussian, Laplace, and randomized response mechanisms, so it can be used as a privacy accountant that only needs alpha, epsilon, sensitivity, and noise scale.
  • The adaptive composition rule lets practitioners add up ADP costs across iterations with the closed-form update; the paper's simulations show the cumulative cost stays below RDP and zCDP for small to moderate iteration counts.
  • Under very strict failure probabilities ($\delta=10^{-25}$), the paper reports that ADP's cumulative privacy consumption grows more slowly than RDP's, making it a candidate for high-sensitivity settings like healthcare and finance.
  • The alpha-selection algorithms give a practical recipe: search alpha in a modest range (roughly 2 to 300) to minimize either cumulative privacy consumption or the required noise scale.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the identity $D_\alpha^{\text{RDP}} = \frac{1}{\alpha-1}\log(1+\alpha(\alpha-1)eD_\alpha)$ converts the paper's Gaussian ADP bound into the standard RDP Gaussian bound, the practical claim reduces to an accounting convention: the same privacy loss can be reported in two currencies, and the reported advantage comes from choosing the currency's alpha to minimize the total.
  • The per-iteration update $\epsilon_{\text{new}} \leftarrow \epsilon + \alpha(\alpha-1)\epsilon\,\epsilon_{\text{new}}$ resembles a moments accountant; a natural extension is a lightweight privacy-budget tracker that records $eD_\alpha$ instead of Rényi divergence, which could be useful in federated learning loops.
  • The group-privacy rescaling of alpha suggests a testable hypothesis: for small groups, ADP's group bound may be tighter than applying standard composition over group members, an experiment one could run on synthetic query workloads.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper introduces alpha differential privacy (ADP), defined by requiring eD_alpha(M(D)||M(D')) <= epsilon for adjacent datasets, where eD_alpha is the alpha divergence of Definition 7. It claims that ADP is a new, more flexible privacy accounting framework that achieves lower cumulative privacy consumption than RDP and zCDP in small-to-moderate iteration regimes, especially under very small failure probabilities. The paper proves post-processing invariance, adaptive sequential composition, group privacy, and a conversion to approximate differential privacy, gives ADP bounds for randomized response, Laplace, and Gaussian mechanisms, and presents simulations comparing ADP with RDP, zCDP, and advanced composition.

Significance. If the main claim were correct, ADP would be a new privacy accounting method with practical advantages for short iterative analyses under strict delta requirements. Some technical calculations in the paper are correct: the randomized-response, Laplace, and Gaussian mechanism bounds in Propositions 5-7 follow from direct integration, and Lemma 1 is a standard verification of the limiting behavior of Renyi divergence. However, the central contribution is not supported because ADP is a monotone reparameterization of RDP: from Definitions 5 and 7, D_alpha^RDP = (1/(alpha-1)) log(1 + alpha(alpha-1) eD_alpha), so (alpha,epsilon)-ADP is exactly (alpha,rho)-RDP with rho = log(1+alpha(alpha-1)epsilon)/(alpha-1). The Gaussian bound in Proposition 7 then reduces to the standard RDP Gaussian bound, and the claimed advantages in Figures 6-8 are artifacts of separately optimizing alpha in a reparameterized curve rather than new privacy guarantees. In addition, the adaptive composition proof contains an invalid factorization, the group privacy proof uses an unproved inequality, and the reported simulation numbers do not match the paper's own formulas.

major comments (4)
  1. [Section 4, Definitions 5-8, Proposition 7, Algorithm 1, Figures 6-8] ADP is a monotone reparameterization of RDP, so the claimed new privacy framework is not supported. From Definition 5, D_alpha^RDP(P||Q) = (1/(alpha-1)) log S with S = integral (dP/dQ)^alpha dQ, while Definition 7 gives eD_alpha(P||Q) = (S-1)/(alpha(alpha-1)). Hence D_alpha^RDP(P||Q) = (1/(alpha-1)) log(1 + alpha(alpha-1) eD_alpha(P||Q)), a strictly increasing bijection. Therefore (alpha,epsilon)-ADP is exactly (alpha,rho)-RDP with rho = log(1+alpha(alpha-1)epsilon)/(alpha-1). Applying this identity to Proposition 7 yields rho = alpha Delta_f^2/(2 sigma_G^2), which is precisely the standard RDP Gaussian bound in Eq. (34), and the conversion in Proposition 4 coincides with the RDP conversion in Eq. (35). Consequently, the lower intercept of ADP in Figures 6-8 is an artifact of optimizing alpha in this reparameterized curve, not evidence of a new guarantee; this invalidates the central claim of the abstract and Section 8.3.
  2. [Proposition 2, Eq. (53)] The proof of adaptive sequential composition is invalid. The third equality in Eq. (53) rewrites the integral over A x B as a product of an integral over A of (dP_X/dP_X')^alpha dP_X' and an integral over B of (dP_{Y|X}/dP_{Y'|X'})^alpha dP_{Y'|X'}. This factorization requires the conditional distribution P_{Y|X} to be independent of X, which is exactly the nonadaptive case; in the adaptive composition considered in the proposition, P_{Y|X} depends on X, so the two factors cannot be separated into a product of integrals. Since composition is the basis for Algorithm 1 and for the reported cumulative privacy consumption, this error is load-bearing.
  3. [Proposition 3, Eqs. (57)-(59)] The group privacy proof uses an unproved inequality. After applying Jensen's inequality, the proof obtains I_alpha(P||Q) <= I_{2alpha}(P||R)^{1/2} I_{2alpha}(R||Q)^{(2alpha-1)/(4alpha)}, and then asserts that this is <= I_{2alpha}(P||R). For the last step one would need I_{2alpha}(R||Q)^{(2alpha-1)/(2alpha)} <= I_{2alpha}(P||R), which is not established and does not follow from the nonnegativity of alpha divergence or from the absolute-continuity chain stated in the lemma. Without this step, the claimed group privacy guarantee does not follow from Lemma 3.
  4. [Section 8.1, Proposition 7] The reported single-query ADP epsilon values in Section 8.1 are inconsistent with Proposition 7. For Figure 6 with sigma_G = 100 and Delta_f = 1, the stated optimal ADP alpha values are 136, 152, and 164; Proposition 7 gives eD_alpha = (1/(alpha(alpha-1)))(exp((alpha^2-alpha)/(2*100^2))-1), which evaluates to approximately 8.2e-5, 9.4e-5, and 1.05e-4, respectively, not 5.00e-5, 5.05e-5, and 5.24e-5 as stated. This discrepancy indicates that the simulation curves are not computed from the paper's stated formulas, undermining the empirical comparison that motivates ADP.
minor comments (4)
  1. [Proposition 4 and Algorithm 1] The conversion formula is presented ambiguously: Eq. (64) and Algorithm 1 line 12 should read bar_epsilon >= (log(1+alpha(alpha-1)epsilon) + log(1/delta))/(alpha-1), not a formula with the logarithm of delta placed in the denominator.
  2. [Definition 7] The f-function in Definition 7 should be typeset as f(u) = (u^alpha - alpha u - (1-alpha))/(alpha(alpha-1)); as printed, 'ualpha-alpha u-(1-alpha)' is ambiguous and could be misread.
  3. [Lemma 3 and Remark 3] The absolute-continuity chain is written as lambda << P << Q << R << lambda, which reverses the usual direction for densities with respect to Lebesgue measure; it should state that P, Q, and R are mutually absolutely continuous and dominated by lambda.
  4. [Section 6.4 and Definition 1] There are several editorial errors, including Definition 1's incomplete phrase 'epsilon-differential is defined as' and the Section 8.2 text referring to colored lines without stating the line styles in the captions; these should be corrected for clarity.

Circularity Check

1 steps flagged · score 8.0 of 10

ADP is a monotone reparameterization of RDP via the identity linking Definitions 5 and 7, so the claimed small-iteration advantage is not a new privacy framework; the central claim is unsupported.

  1. renaming known result [Definition 5 (13), Definition 7 (42), Definition 8 (47), Proposition 7 (74), Algorithm 1, Section 8.2 and 8.3]
    "Dα(P∥Q) = 1/(α−1) log(∫ X (dP/dQ)^α dQ) ... eDα(P∥Q)≜ 1/(α(α−1))(∫ X pαq1−αdλ−1) ... The main observation is that although the privacy consumption of both ADP and RDP estimates shows a linear growth, the lower intercept of ADP shows that it provides a stronger initial privacy estimate."

    By the paper's own Definitions 5 and 7, writing S=∫p^α q^{1−α}dλ gives Dα=log S/(α−1) and eDα=(S−1)/(α(α−1)). Hence (α,ε)-ADP is exactly (α,ρ)-RDP with ρ=log(1+α(α−1)ε)/(α−1), a strictly increasing bijection. Substituting Proposition 7's Gaussian ADP value yields ρ=αΔf₂²/(2σ_G²), the standard RDP Gaussian bound, and Algorithm 1's recurrence/conversion reduces to T-fold RDP composition with the standard conversion. The lower ADP intercept claimed in Section 8.3 is therefore not an independent privacy guarantee; it is an artifact of plotting the same RDP accounting under a reparameterization, so the central comparison of ADP against RDP is circular.

full rationale

The paper's load-bearing claim is that (α,ε)-alpha differential privacy (Definition 8) is a new framework that can beat RDP in small-to-moderate iteration regimes. That claim collapses under the paper's own definitions: alpha divergence and Rényi divergence are computed from the same integral S=∫p^α q^{1−α}dλ, related by the increasing bijection Dα=log(1+α(α−1)eDα)/(α−1). Consequently the ADP condition eDα≤ε is equivalent to the RDP condition Dα≤ρ with ρ the corresponding transformed value; ADP post-processing, composition, and conversion to (ε,δ)-DP are RDP post-processing, composition, and conversion in disguise. Proposition 7's Gaussian bound transforms to the textbook RDP Gaussian bound, and Algorithm 1 optimizes the same objective as RDP would. The 'lower intercept' and strict-δ advantages highlighted in Figures 6-8 and Section 8.3 are therefore forced by the reparameterization and by separately optimizing α in equivalent equations, not by any new privacy accounting; this is a renaming of a known result rather than a novel framework. I did not count any self-citations as load-bearing: the circularity is internal to the paper's equations. Separately, Proposition 2's adaptive-composition proof contains an invalid factorization (the third equality of Eq. 53 treats P_Y|X as independent of X), and the group-privacy proof's step from Eq. 58 to Eq. 59 is unsupported; those are correctness defects independent of the circularity finding.

Assumptions & free parameters 2 free parameters · 4 assumptions · 1 invented entities

The central claim rests on treating alpha divergence as a new privacy metric, plus a set of analytic inequalities. The alpha values in the simulations are optimized separately for each framework, which drives the claimed advantage. The composition and group privacy proofs depend on unproved or false structural assumptions. No physical entities are introduced; the only invented object is the ADP definition itself.

free parameters (2)
  • alpha_ADP = 136, 152, 164 (Figure 6); 13, 64, 127 (Figure 7); 14, 67, 133 (Figure 8)
    User-searched to minimize cumulative privacy consumption for the ADP curve. The claimed advantage is sensitive to this optimization and is not compared at matched alpha with RDP.
  • alpha_RDP = 69, 97, 119 (Figure 6); 6, 25, 49 (Figure 7); 7, 32, 63 (Figure 8)
    The comparison optimizes alpha separately for RDP. Because ADP and RDP are equivalent, the different optima are numerical artifacts of the same objective, not evidence of a different framework.
assumptions (4)
  • ad hoc to paper Joint density ratio factorization in adaptive composition (Eq. 53)
    The equality in Eq. (53) treats the conditional density ratio as independent of the conditioning variable, which is false for adaptive mechanisms; the proof would need an extra inequality.
  • ad hoc to paper Inequality I_{2alpha}(R||Q)^((2alpha-1)/(4alpha)) <= I_{2alpha}(P||R)^(1/2) in group privacy proof
    Used at Eq. (59) without proof; not entailed by the stated ADP conditions, so the group privacy bound is not established as written.
  • standard math Jensen, Holder, Markov, and L'Hopital are applied
    These tools are standard and mostly applied correctly in the individual mechanism derivations.
  • domain assumption Absolute continuity chain lambda << P << Q << R << lambda for adjacent mechanism outputs
    Required by Lemma 3; Remark 3 argues commonly used mechanisms satisfy it, but this is an additional regularity assumption not proven for all ADP mechanisms.
invented entities (1)
  • Alpha differential privacy (ADP)
    purpose: New privacy definition replacing the max-divergence epsilon bound with a bound on alpha divergence.
    Equivalent to Renyi differential privacy under D_alpha^RDP = (1/(alpha-1)) log(1+alpha(alpha-1)eD_alpha); no independent falsifiable content beyond the existing RDP and f-DP frameworks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Novel Approach to Differential Privacy with Alpha Divergence." pith.science (2026). https://pith.science/paper/FM53EZBS

@misc{pith2026250617012,
  author       = {Pith},
  title        = {Pith review of: A Novel Approach to Differential Privacy with Alpha Divergence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FM53EZBS}},
  note         = {Machine review of arXiv:2506.17012}
}
abstract

As data-driven technologies advance swiftly, maintaining strong privacy measures becomes progressively difficult. Conventional $(\epsilon, \delta)$-differential privacy, while prevalent, exhibits limited adaptability for many applications. To mitigate these constraints, we present alpha differential privacy (ADP), an innovative privacy framework grounded in alpha divergence, which provides a more flexible assessment of privacy consumption. This study delineates the theoretical underpinnings of ADP and contrasts its performance with competing privacy frameworks across many scenarios. Empirical assessments demonstrate that ADP offers enhanced privacy guarantees in small to moderate iteration contexts, particularly where severe privacy requirements are necessary. The suggested method markedly improves privacy-preserving methods, providing a flexible solution for contemporary data analysis issues in a data-centric environment.

Figures

Figures reproduced from arXiv: 2506.17012 by the authors.

Figure 1
Figure 1. privacy consumption of the randomized response mechanism under different probabilities [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. privacy consumption of the Laplace mechanism under different scales [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. privacy consumption of the Gaussian mechanism under different standard deviation [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Cumulative privacy consumption (ϵ) of the Gaussian mechanism over 1000 iterations under varying required standard deviations, with a required failure rate of δ =1e-5 and a required sensitivity of ∆f2 = 1. The horizontal axis represents the value of α, while the vertica…
Figure 5
Figure 5. Figure 5: Optimal standard deviation (σG) of the Gaussian mechanism over 1000 iterations under varying required upper bounds of cumulative privacy consumption, with a required failure rate of δ = 1e-5 and a required sensitivity of ∆f2 = 1. The horizontal axis represents the valu…
Figure 6
Figure 6. Figure 6: Comparison of cumulative privacy consumption between alpha differential privacy and other mainstream [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Comparison of cumulative privacy consumption between alpha differential privacy and other mainstream [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]
Figure 8
Figure 8. Figure 8: Comparison of cumulative privacy consumption between alpha differential privacy and other mainstream [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 17 canonical work pages

  1. [1]

    Calibrating noise to sensitivity in private data analysis,

    C. Dwork, F. McSherry, K. Nissim, and A. Smith, “Calibrating noise to sensitivity in private data analysis," in Theory of Cryptography Conference, 2006, pp. 265-284

    work page 2006

  2. [2]

    Differential privacy,

    C. Dwork, "Differential privacy," inInternational Colloquium on Automata, Languages, and Programming, 2006, pp. 1-12

    work page 2006

  3. [3]

    Deep learning with differential privacy,

    M. Abadi, A. Chu, I. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang, “Deep learning with differential privacy," inProceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security, 2016, pp. 308-318

    work page 2016

  4. [4]

    Collecting and analyzing multidimen- sional data with local differential privacy,

    N. Wang, X. Xiao, Y . Yang, J. Zhao, S. C. Hui, H. Shin, J. Shin, and G. Yu, “Collecting and analyzing multidimen- sional data with local differential privacy," in2019 IEEE 35th International Conference on Data Engineering (ICDE), 2019, pp. 638-649

    work page 2019

  5. [5]

    Data mining with differential privacy,

    A. Friedman and A. Schuster, “Data mining with differential privacy," inProceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2010, pp. 493-502

    work page 2010

  6. [6]

    Improving the Gaussian mechanism for differential privacy: Analytical calibration and optimal denoising,

    B. Balle and Y .-X. Wang, “Improving the Gaussian mechanism for differential privacy: Analytical calibration and optimal denoising," inInternational Conference on Machine Learning, 2018, pp. 403-412

    work page 2018

  7. [7]

    The composition theorem for differential privacy,

    P. Kairouz, S. Oh, and P. Viswanath, “The composition theorem for differential privacy,"IEEE Transactions on Information Theory, vol. 61, no. 6, pp. 3391-3406, 2015

    work page 2015

  8. [8]

    Amari,Information Geometry and Its Applications

    S. Amari,Information Geometry and Its Applications. Springer, 2016

    work page 2016

Show all 19 references
  1. [9]

    Information theory and statistics: A tutorial,

    I. Csiszár and P. C. Shields, “Information theory and statistics: A tutorial,"Foundations and Trends® in Communi- cations and Information Theory, vol. 1, no. 4, pp. 417-528, 2004

    2004

  2. [10]

    The Algorithmic Foundations of Differential Privacy,

    C. Dwork and A. Roth, “The Algorithmic Foundations of Differential Privacy,"Foundations and Trends® in Theoretical Computer Science, vol. 9, no. 3-4, pp. 211-407, 2014

    2014

  3. [11]

    Our data, ourselves: Privacy via distributed noise generation,

    C. Dwork, K. Kenthapadi, F. McSherry, I. Mironov, and M. Naor, “Our data, ourselves: Privacy via distributed noise generation," inAdvances in Cryptology-EUROCRYPT 2006, St. Petersburg, Russia, 2006, pp. 486-503, Springer

    2006

  4. [12]

    Rényi differential privacy,

    I. Mironov, “Rényi differential privacy," in2017 IEEE 30th Computer Security Foundations Symposium (CSF), 2017, pp. 263-275

    2017

  5. [13]

    Concentrated differential privacy: Simplifications, extensions, and lower bounds,

    M. Bun and T. Steinke, “Concentrated differential privacy: Simplifications, extensions, and lower bounds," in Theory of Cryptography Conference, 2016, pp. 635-658, Springer

    2016

  6. [14]

    Non-negative matrix factorization withα-divergence,

    A. Cichocki, H. Lee, Y .-D. Kim, and S. Choi, “Non-negative matrix factorization withα-divergence,"Pattern Recognition Letters, vol. 29, no. 9, pp. 1433–1440, 2008

    2008

  7. [15]

    Mathematical aspects of divergence based vector quantization using Fréchet- derivatives,

    T. Villmann and S. Haase, “Mathematical aspects of divergence based vector quantization using Fréchet- derivatives,"University of Applied Sciences Mittweida, 2010

    2010

  8. [16]

    Information-type measures of difference of probability distributions and indirect observations,

    I. Csiszár, “Information-type measures of difference of probability distributions and indirect observations,"Studia Scientiarum Mathematicarum Hungarica, vol. 2, pp. 299-318, 1967

    1967

  9. [17]

    Rényi divergence and Kullback-Leibler divergence,

    T. van Erven and P. Harremoës, “Rényi divergence and Kullback-Leibler divergence,"IEEE Transactions on Information Theory, vol. 60, no. 7, pp. 3797-3820, Jul. 2014. [Online]. Available: arxiv.org/abs/1206.2459. 22 A Novel Approach to Differential Privacy with Alpha DivergenceA...

    2014 arXiv

  10. [18]

    Privacy integrated queries: An extensible platform for privacy-preserving data analysis,

    F. D. McSherry, “Privacy integrated queries: An extensible platform for privacy-preserving data analysis," in Proceedings of the 2009 ACM SIGMOD International Conference on Management of Data, C. Binnig and B. Dageville, Eds., 2009, pp. 19-30

    2009

  11. [19]

    Boosting and differential privacy,

    C. Dwork, G. N. Rothblum, and S. Vadhan, “Boosting and differential privacy," in2010 IEEE 51st Annual Symposium on Foundations of Computer Science (FOCS), 2010, pp. 51–60. doi: 10.1109/FOCS.2010.12. 23

    2010 doi

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.