Pith. sign in

REVIEW 3 major objections 5 minor 44 references

Privacy-Preserving Federated Convex Optimization: Balancing Partial-Participation and Efficiency via Noise Cancellation

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims a noise-cancellation mechanism achieves optimal excess population loss for differentially private federated convex optimization under partial participation, matching known lower bounds while using only a linear number of…

desk verdict A genuine new mechanism for partial-participation DP-FL with an optimal-looking rate, but the main proof has a gap in the noise-correlation term that needs closing before the claim is solid. read the letter →

arxiv 2506.02563 v1 pith:5B4OXWWI submitted 2025-06-03 cs.LG

Roie Reshef , Kfir Yehuda Levy This is my paper
classification cs.LG
keywords differentialprivacyfederatedlearningpartialparticipationstochasticconvexoptimizationnoisecancellationuntrustedserverlinear-timepopulationloss
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

This paper claims that private federated learning can have it both ways: even when only a random subset of machines participates in each round, and even when the server cannot be trusted with raw gradients, a differentially private algorithm can reach the optimal population-error rate—the best rate allowed by known lower bounds—for both homogeneous and heterogeneous data distributions, while using only a linear number of gradient computations. The obstacle is that a machine sitting out a round still holds a noise value the server's aggregate depends on, making the injected noise correlated across rounds. The paper's noise-cancellation scheme addresses this: each participating machine adds fresh noise and subtracts its previous noise, so the server's cumulative view is one noise per machine normalized by the participating subset size. The stated rates are O(1/√n + √d/(ε n)) for a trusted server and O(1/√n + √(M d)/(ε n)) for an untrusted server, matching established lower bounds while using two gradient evaluations per sample.

What carries the argument

The load-bearing object is the noise-cancellation mechanism combined with the µ2-SGD double-momentum update. Each machine i that participates at round t broadcasts s_{t,i} = α_t ∇f(x_t; z_{t,i}) − α_{t−1}∇f(x_{t−1}; z_{t,i}) plus the difference Y_{t,i} − Y_{t−1,i} of fresh and previous Gaussian noise; machines that do not participate keep their previous noise Y_{t,i} = Y_{t−1,i}. Telescoping makes the server's noisy cumulative estimate satisfy q̃_t = q_t + (1/m) Σ_{i=1}^M Y_{t,i}, so the effective injected noise is a sum of M independent noises normalized by m, rather than growing with the number of rounds participated. The variance σ²_{t,i} of each fresh noise is scaled by N_{t,i}, the number of rounds machine i has participated so far; the harmonic sum Σ 1/N_{t,i} ≤ 1+log T makes the total privacy budget (ρ²/2)-zCDP independent of the participation pattern, while the convergence proof bounds the correlated-noise term using a geometric-weight argument with participation probability p = m/M.

What would settle it

Run Algorithm 1 on a synthetic convex problem with M=2 machines, m=1 participant per round, identical data, and compare independent fair-coin participation against deterministic alternation (machine 1 on odd rounds, machine 2 on even rounds). The proof's bound on the correlated-noise term assumes a Bernoulli participation process, so the alternating schedule invalidates that step; if the measured excess loss under alternation exceeds the uniform-schedule loss by an amount that grows with the number of rounds, the uniformity assumption is essential to the claimed rate.

Watch

Extended reading notes

Core claim

Under convex, G-Lipschitz, and L-smooth losses with i.i.d. samples per machine, Algorithm 1 is claimed to be (ρ²/2)-zCDP (a concentrated differential privacy guarantee) for each machine and, after T rounds with m participants each, to have expected excess population loss whose order is O(1/√n + √(M d)/(ε n)) for the untrusted-server case and O(1/√n + √d/(ε n)) for the trusted-server case, where n=mT is the total number of samples used and ε is the converted DP level. These match the known lower bounds for the two settings. Each participating machine uses one fresh sample per round and performs two gradient evaluations on it, so the total number of gradient computations is linear in n, the same order as ordinary non-private SGD. The central assertion is that privacy, partial participation, optimal statistical error, and linear-time computation are simultaneously attainable.

Load-bearing premise

The load-bearing premise is that the subset of machines participating in each round is chosen uniformly at random from all subsets of size m, independently across rounds; if device availability is data-dependent, adversarial, or correlated over time, the proof's calculation of how the accumulated noise grows stops working and the claimed optimal rate is not established.

Editorial extensions

If this is right

  • Partial participation no longer forces a multiplicative √(M/m) slowdown: the untrusted-server excess loss matches the known lower bound for the setting.
  • A deployment with M machines and m participants per round can train with the same order of gradient computations as standard SGD while keeping per-machine differential privacy against an untrusted server.
  • The trusted-server variant reaches the plain DP-SCO lower bound with the same linear-time budget, so the extra √M privacy penalty is tied to hiding data from the server, not to partial participation.
  • Because each sample is used once, the algorithm is single-pass; it avoids the mega-batch resampling that made prior optimal methods require O(n^{3/2}) or O(n^{9/8}) work.
  • Heterogeneous data distributions are absorbed through the bounded-heterogeneity parameters, so the optimal rate does not require identical machines.

Reading between the lines

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

  • The noise-cancellation structure is presented for convex SCO; a natural extension would be to apply the same add-fresh-subtract-last noise update to non-convex losses or adaptive-gradient methods, where the sensitivity bound on the correction term would need to be rederived.
  • The uniform-random-participation assumption could likely be relaxed to any schedule with known participation probabilities, but the proof's geometric-weight bound degrades with the schedule's autocorrelation; device-availability logs could be used to measure how far a real deployment is from the Bernoulli ideal.
  • The method's single-pass property suggests it fits settings where on-device data cannot be stored or revisited; a practical stress test would run the algorithm with participation driven by real device-availability traces and compare against the bound.
  • If the lower bound is tight, any further improvement in this model would have to change the threat model or the communication model—for example secure aggregation or trusted hardware—rather than the optimization algorithm.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper studies differentially private federated stochastic convex optimization under partial participation. It proposes Algorithm 1, a variant of the μ2-SGD technique in which each participating machine sends a corrected gradient increment plus a 'last noise' cancellation term, and the server uses the aggregate to update a weighted iterate. The stated contributions are: a trusted-server algorithm with rate O(1/√n + √d/(ε n)) and linear gradient complexity; an untrusted-server algorithm with rate O(1/√n + √(Md)/(ε n)), matching the lower bound of Lowy and Razaviyayn (2023); and a noise scheduling rule σ²_{t,i} ∝ N_{t,i} proportional to the machine's participation count. The appendix contains proofs of privacy and convergence, and the experiments report MNIST logistic-regression results comparing the new method with Noisy SGD and with Lowy-Razaviyayn.

Significance. If the central claim were fully proven, this would be a substantial advance: an optimal excess population loss with O(n) gradient computations in the untrusted-server partial-participation setting, improving on the n^{3/2} and n^{9/8} methods of Lowy-Razaviyayn and Gao et al. The noise-cancellation mechanism and the participation-dependent noise schedule are natural and interesting ideas, and the claimed rates match known lower bounds. However, the main convergence theorem (Theorem 5.2) is not established by the proof as written because of an invalid correlation step in the handling of term (C); the appendix also contains an incorrect proof of a standard composition lemma. The paper is therefore not yet publishable in its current form.

major comments (3)
  1. [Appendix E.2, term (C)] The derivation of term (C) replaces the indicator that machine i's last participation before time τ was at time s with its marginal probability p(1−p)^{τ−s} inside the expectation E⟨y_{s,i}, x*−w_{τ+1}⟩. This step is not justified: w_{τ+1} is a function of the participation indicators and of y_{s,i} whenever that indicator is one, so the event and the iterate are dependent. The geometric weighting is what produces the m in the denominator of the final privacy-related term in Theorem 5.2, so this is not a cosmetic issue. A valid proof of Theorem 5.2 needs either a martingale or conditional-expectation argument that keeps the indicator inside the expectation, or a different bound on ∑_{τ=1}^{t} E⟨Y_τ, x*−w_{τ+1}⟩.
  2. [Appendix A.3, proof of Lemma A.3] The induction step in the proof of the composition lemma splits log E[XY] as log E[X] + log E[Y] for the two likelihood-ratio factors. This identity would require X and Y to be independent, which is not established; conditional independence of A_{k+1} given the previous outputs does not make the two factors independent under the joint expectation. The lemma itself is standard, so the fix is easy (cite Mironov or Bun–Steinke), but as printed the proof is incorrect and should be replaced by a correct proof or a citation.
  3. [Appendix E.2, final bounding step after Theorem A.7] The proof uses the inequality 1/p = M/m ≤ T to replace T + 1/p by O(T). This inequality is not among the hypotheses of Theorem 5.2 and can fail when the number of machines exceeds the total number of samples used (for example, M = 10^6, m = 100, T = 100). Without that assumption the final privacy term carries an extra M/(mT) factor and the claimed optimal rate is not obtained. The theorem should either state M ≤ mT explicitly or carry the exact dependence in the bound.
minor comments (5)
  1. [Section 5.1 and Appendix E.1] The text repeatedly refers to 'Theorem 4.2' when the displayed statement in Section 4.3 is labeled 'Lemma 4.2'; please unify the numbering across the main text and appendix.
  2. [Theorem 5.2 statement] The uniform i.i.d. participation model is essential for the proof, but the abstract and introduction present the result as applying to partial participation generally. Please state prominently that the optimality claim is for the uniform, independent-per-round participation model and not for adversarial or correlated device availability.
  3. [Algorithm 1 input description] Algorithm 1 lists a dataset {z_{1,i},…,z_{T,i}} for every machine i, while the introduction says the total number of samples used is n = mT. Please clarify that unselected machines' samples are not used in training and explain the relationship between M, m, T, and the total data size n.
  4. [Section 5.3, experiments] The comparison with 'Other Work' is not on equal footing because that baseline reuses samples while the proposed method makes a single pass over the data; the text acknowledges this but should state it as a limitation of the experimental comparison before drawing conclusions about accuracy.
  5. [Throughout] The claimed rates O(1/√n + √(Md)/(ε n)) are stated without mentioning the logarithmic factor (1+log T) that appears in Theorem 5.2. Please use O-tilde notation or explicitly state that the matching to the lower bound is up to logarithmic factors.

Circularity Check

0 steps flagged · score 1.0 of 10

No construction-level circularity: the partial-participation rate is derived from a new correlated-noise analysis and checked against external lower bounds; reliance on the authors' prior µ2-SGD framework is prior work, not an input/output identification.

full rationale

I walked the claimed derivation chain. The headline claim (Theorem 5.2) is an upper bound on the excess loss of Algorithm 1. The proof decomposes the regret into gradient terms, the momentum error ε_t bounded by Lemma 4.3, and the correlated noise term (C) = Σ_t E⟨Y_t, x*−w_{t+1}⟩. Term (C) is the genuinely new part: the paper sets σ²_{t,i}=4S²(1+logT)/ρ²·N_{t,i} and then bounds Σ E⟨Y_τ, x*−w_{τ+1}⟩ directly in Appendix E.2. This is not a fitted constant renamed as a prediction: the learning rate η and the noise schedule are analytic functions of the problem parameters, and the final √(Md)/(ρmT) term emerges from summing geometric weights over the M machines. The optimality comparison is made against external lower bounds, Bassily et al. (2014) and Lowy & Razaviyayn (2023), not against quantities fitted inside this paper. The self-citations to Levy (2023) and Reshef & Levy (2024) supply the µ2-SGD optimizer and the full-participation DP analysis; those are published prior derivations, and the partial-participation noise-cancellation mechanism is analyzed in this paper rather than imported as an identical result. The trusted-server theorem in Appendix F is explicitly delegated: 'The proof is the same as the one in (Reshef & Levy, 2024),' but that delegation does not carry the central untrusted-server partial-participation claim. I also note a genuine proof-rigor concern in Appendix E.2: the step 'Y_{τ,i} is actually y_{s,i}, where s is a geometric random variable... ys,i is independent of everything up to the time step s, so we replace x∗ with ws' inserts the marginal geometric probability inside an expectation containing future correlated iterates; if wrong, the claimed rate is not established. That is a correctness gap, not a circularity, because the claimed rate is not definitionally equal to an input of Algorithm 1 or to a fitted parameter. Accordingly, the score stays in the normal 0–2 range with no specific circular step identified.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard convex SCO assumptions and on a uniform random participation model. The paper introduces no new physical or mathematical entities. Its main burden is the correctness of the adaptive privacy composition argument and the noise correlation analysis, which are not fully rigorous in the written proofs.

assumptions (5)
  • domain assumption Each local objective f_i is convex, G-Lipschitz, and L-smooth over a compact convex set K of diameter D (Section 2.1 Assumptions).
    These assumptions define the SCO setting and are used throughout the proofs of Theorems 4.2, 5.1, and 5.2.
  • domain assumption Each machine i has a dataset S_i of T i.i.d. samples from D_i, and a participating machine draws a fresh sample each round (Algorithm 1 and Section 2.1).
    The unbiased gradient estimates and the total sample count n=mT depend on fresh i.i.d. draws.
  • domain assumption Participation subsets M_t are chosen uniformly from all size-m subsets of [M], independently across rounds (Theorem 5.2 and Section 2.1).
    The noise analysis of term (C) and Lemma 2.1 rely on uniform random subsets with fixed size m.
  • standard math Standard RDP tools: Gaussian mechanism bound, composition, and post-processing (Lemmas 2.3, A.3, A.4).
    These are textbook tools for Renyi differential privacy; the composition lemma is standard, though the paper's proof of it contains an invalid step.
  • domain assumption External lower bounds of Bassily et al. (2014) and Lowy and Razaviyayn (2023) apply to this model with n=mT used samples (Section 2.3).
    The optimality claim identifies the paper's upper bound with these lower bounds; if the lower bounds are not valid for this participation model, the claim of optimality would need re-examination.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Privacy-Preserving Federated Convex Optimization: Balancing Partial-Participation and Efficiency via Noise Cancellation." pith.science (2026). https://pith.science/paper/5B4OXWWI

@misc{pith2026250602563,
  author       = {Pith},
  title        = {Pith review of: Privacy-Preserving Federated Convex Optimization: Balancing Partial-Participation and Efficiency via Noise Cancellation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5B4OXWWI}},
  note         = {Machine review of arXiv:2506.02563}
}
read the original abstract

This paper tackles the challenge of achieving Differential Privacy (DP) in Federated Learning (FL) under partial-participation, where only a subset of the machines participate in each time-step. While previous work achieved optimal performance in full-participation settings, these methods struggled to extend to partial-participation scenarios. Our approach fills this gap by introducing a novel noise-cancellation mechanism that preserves privacy without sacrificing convergence rates or computational efficiency. We analyze our method within the Stochastic Convex Optimization (SCO) framework and show that it delivers optimal performance for both homogeneous and heterogeneous data distributions. This work expands the applicability of DP in FL, offering an efficient and practical solution for privacy-preserving learning in distributed systems with partial participation.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages

  1. [1]

    J., McMahan, H

    Abadi, M., Chu, A., Goodfellow, I. J., McMahan, H. B., Mironov, I., Talwar, K., and Zhang, L. Deep learning with differential privacy. In Weippl, E. R., Katzenbeisser, S., Kruegel, C., Myers, A. C., and Halevi, S. (eds.), Proceedings of ACM SIGSAC Conference on Computer and Communications Security . ACM , 2016

    work page 2016

  2. [2]

    Differentially private inference via noisy optimization

    Avella - Medina, M., Bradshaw, C., and Loh, P. Differentially private inference via noisy optimization. CoRR, abs/2103.11003, 2021

    work page Pith review arXiv 2021

  3. [3]

    D., and Thakurta, A

    Bassily, R., Smith, A. D., and Thakurta, A. Differentially private empirical risk minimization: Efficient algorithms and tight error bounds. In 55th IEEE Annual Symposium on Foundations of Computer Science, FOCS . IEEE Computer Society, 2014

    work page 2014

  4. [4]

    Bassily, R., Feldman, V., Talwar, K., and Thakurta, A. G. Private stochastic convex optimization with optimal rates. In Wallach, H. M., Larochelle, H., Beygelzimer, A., d'Alch \' e - Buc, F., Fox, E. B., and Garnett, R. (eds.), Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems, NeurIPS, 2019

    work page 2019

  5. [5]

    and Steinke, T

    Bun, M. and Steinke, T. Concentrated differential privacy: Simplifications, extensions, and lower bounds. In Hirt, M. and Smith, A. D. (eds.), Theory of Cryptography - 14th International Conference, TCC , Proceedings, Part I , Lecture Notes in Computer Science, 2016

    work page 2016

  6. [6]

    Chaudhuri, K., Monteleoni, C., and Sarwate, A. D. Differentially private empirical risk minimization. J. Mach. Learn. Res., 2011

    work page 2011

  7. [7]

    Shuffle private stochastic convex optimization

    Cheu, A., Joseph, M., Mao, J., and Peng, B. Shuffle private stochastic convex optimization. In The Tenth International Conference on Learning Representations, ICLR , 2022

    work page 2022

  8. [8]

    Anytime online-to-batch, optimism and acceleration

    Cutkosky, A. Anytime online-to-batch, optimism and acceleration. In Chaudhuri, K. and Salakhutdinov, R. (eds.), Proceedings of the 36th International Conference on Machine Learning, ICML , Proceedings of Machine Learning Research. PMLR , 2019

    work page 2019

Show all 44 references
  1. [9]

    and Orabona, F

    Cutkosky, A. and Orabona, F. Momentum-based variance reduction in non-convex SGD . In Wallach, H. M., Larochelle, H., Beygelzimer, A., d'Alch \' e - Buc, F., Fox, E. B., and Garnett, R. (eds.), Advances in Neural Information Processing Systems 32: Annual Conference on Neural I...

    2019

  2. [10]

    Optimal distributed online prediction using mini-batches

    Dekel, O., Gilad - Bachrach, R., Shamir, O., and Xiao, L. Optimal distributed online prediction using mini-batches. J. Mach. Learn. Res., 2012

    2012

  3. [11]

    C., Jordan, M

    Duchi, J. C., Jordan, M. I., and Wainwright, M. J. Local privacy and statistical minimax rates. In 54th Annual IEEE Symposium on Foundations of Computer Science, FOCS . IEEE Computer Society, 2013

    2013

  4. [12]

    and Roth, A

    Dwork, C. and Roth, A. The algorithmic foundations of differential privacy. Found. Trends Theor. Comput. Sci., 2014

    2014

  5. [13]

    Our data, ourselves: Privacy via distributed noise generation

    Dwork, C., Kenthapadi, K., McSherry, F., Mironov, I., and Naor, M. Our data, ourselves: Privacy via distributed noise generation. In Vaudenay, S. (ed.), Advances in Cryptology - EUROCRYPT , 25th Annual International Conference on the Theory and Applications of Cryptographic Te...

    2006

  6. [14]

    Dwork, C., McSherry, F., Nissim, K., and Smith, A. D. Calibrating noise to sensitivity in private data analysis. In Halevi, S. and Rabin, T. (eds.), Theory of Cryptography, Third Theory of Cryptography Conference, TCC , Lecture Notes in Computer Science. Springer, 2006 b

    2006

  7. [15]

    Private stochastic convex optimization: optimal rates in linear time

    Feldman, V., Koren, T., and Talwar, K. Private stochastic convex optimization: optimal rates in linear time. In Makarychev, K., Makarychev, Y., Tulsiani, M., Kamath, G., and Chuzhoy, J. (eds.), Proccedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing, STOC . ...

    2020

  8. [16]

    Faster differentially private convex optimization via second-order methods

    Ganesh, A., Haghifam, M., Steinke, T., and Thakurta, A. Faster differentially private convex optimization via second-order methods. CoRR, abs/2305.13209, 2023

    2023 arXiv

  9. [17]

    Gao, C., Lowy, A., Zhou, X., and Wright, S. J. Private heterogeneous federated learning without a trusted server revisited: Error-optimal and communication-efficient algorithms for convex losses. In Forty-first International Conference on Machine Learning, ICML , 2024

    2024

  10. [18]

    M., Data, D., Diggavi, S

    Girgis, A. M., Data, D., Diggavi, S. N., Kairouz, P., and Suresh, A. T. Shuffled model of federated learning: Privacy, accuracy and communication trade-offs. IEEE J. Sel. Areas Inf. Theory , 2021

    2021

  11. [19]

    H., and O'Shea, D

    Hafeez, K., Rehmani, M. H., and O'Shea, D. DPNCT: A differential private noise cancellation scheme for load monitoring and billing for smart meters. In IEEE International Conference on Communications Workshops, ICC Workshops . IEEE , 2021

    2021

  12. [20]

    DP-ADMM: admm-based distributed learning with differential privacy

    Huang, Z., Hu, R., Guo, Y., Chan - Tin, E., and Gong, Y. DP-ADMM: admm-based distributed learning with differential privacy. IEEE Trans. Inf. Forensics Secur. , 2020

    2020

  13. [21]

    P., Song, D., Thakkar, O., Thakurta, A., and Wang, L

    Iyengar, R., Near, J. P., Song, D., Thakkar, O., Thakurta, A., and Wang, L. Towards practical differentially private convex optimization. In IEEE Symposium on Security and Privacy, SP . IEEE , 2019

    2019

  14. [22]

    Practical and private (deep) learning without sampling or shuffling

    Kairouz, P., McMahan, B., Song, S., Thakkar, O., Thakurta, A., and Xu, Z. Practical and private (deep) learning without sampling or shuffling. In Meila, M. and Zhang, T. (eds.), Proceedings of the 38th International Conference on Machine Learning, ICML , Proceedings of Machine...

    2021

  15. [23]

    B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A

    Kairouz, P., McMahan, H. B., Avent, B., Bellet, A., Bennis, M., Bhagoji, A. N., Bonawitz, K. A., Charles, Z., Cormode, G., Cummings, R., D'Oliveira, R. G. L., Eichner, H., Rouayheb, S. E., Evans, D., Gardner, J., Garrett, Z., Gasc \' o n, A., Ghazi, B., Gibbons, P. B., Grutese...

    2021

  16. [24]

    P., Lee, H

    Kasiviswanathan, S. P., Lee, H. K., Nissim, K., Raskhodnikova, S., and Smith, A. D. What can we learn privately? SIAM J. Comput. , 2011

    2011

  17. [25]

    Y., Bach, F

    Kavis, A., Levy, K. Y., Bach, F. R., and Cevher, V. Unixgrad: A universal, adaptive algorithm with optimal guarantees for constrained optimization. In Wallach, H. M., Larochelle, H., Beygelzimer, A., d'Alch \' e - Buc, F., Fox, E. B., and Garnett, R. (eds.), Advances in Neural...

    2019

  18. [26]

    D., and Thakurta, A

    Kifer, D., Smith, A. D., and Thakurta, A. Private convex optimization for empirical risk minimization with applications to high-dimensional regression. In Mannor, S., Srebro, N., and Williamson, R. C. (eds.), COLT 2012 - The 25th Annual Conference on Learning Theory , JMLR Pro...

    2012

  19. [27]

    K., and McMahan, H

    Koloskova, A., McKenna, R., Charles, Z., Rush, J. K., and McMahan, H. B. Gradient descent with linearly correlated noise: Theory and applications to differential privacy. In Advances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processin...

    2023

  20. [28]

    Levy, K. Y. \( \) \( ^ 2 \)-sgd: Stable stochastic optimization via a double momentum mechanism. CoRR, abs/2304.04172, 2023

    2023 arXiv

  21. [29]

    and Razaviyayn, M

    Lowy, A. and Razaviyayn, M. Private federated learning without a trusted server: Optimal algorithms for convex losses. In The Eleventh International Conference on Learning Representations, ICLR , 2023

    2023

  22. [30]

    Private non-convex federated learning without a trusted server

    Lowy, A., Ghafelebashi, A., and Razaviyayn, M. Private non-convex federated learning without a trusted server. In Ruiz, F. J. R., Dy, J. G., and van de Meent, J. (eds.), International Conference on Artificial Intelligence and Statistics, Proceedings of Machine Learning Researc...

    2023

  23. [31]

    McMahan, B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. Communication-efficient learning of deep networks from decentralized data. In Singh, A. and Zhu, X. J. (eds.), Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, AISTAT...

    2017

  24. [32]

    R \' e nyi differential privacy

    Mironov, I. R \' e nyi differential privacy. In 30th IEEE Computer Security Foundations Symposium, CSF . IEEE Computer Society, 2017

    2017

  25. [33]

    Differentially private federated learning on heterogeneous data

    Noble, M., Bellet, A., and Dieuleveut, A. Differentially private federated learning on heterogeneous data. In Camps - Valls, G., Ruiz, F. J. R., and Valera, I. (eds.), International Conference on Artificial Intelligence and Statistics, AISTATS , Proceedings of Machine Learning...

    2022

  26. [34]

    On measures of entropy and information

    R \' e nyi, A. On measures of entropy and information. In Proceedings of the fourth Berkeley symposium on mathematical statistics and probability. The Regents of the University of California, 1961

    1961

  27. [35]

    and Levy, K

    Reshef, R. and Levy, K. Y. Private and federated stochastic convex optimization: Efficient strategies for centralized systems. In Forty-first International Conference on Machine Learning, ICML , 2024

    2024

  28. [36]

    Stochastic convex optimization

    Shalev - Shwartz, S., Shamir, O., Srebro, N., and Sridharan, K. Stochastic convex optimization. In COLT - The 22nd Conference on Learning Theory , 2009

    2009

  29. [37]

    Song, S., Chaudhuri, K., and Sarwate, A. D. Stochastic gradient descent with differentially private updates. In IEEE Global Conference on Signal and Information Processing, GlobalSIP . IEEE , 2013

    2013

  30. [38]

    Nearly optimal private LASSO

    Talwar, K., Thakurta, A., and Zhang, L. Nearly optimal private LASSO . In Cortes, C., Lawrence, N. D., Lee, D. D., Sugiyama, M., and Garnett, R. (eds.), Advances in Neural Information Processing Systems 28: Annual Conference on Neural Information Processing Systems, 2015

    2015

  31. [39]

    and Smith, A

    Thakurta, A. and Smith, A. D. Differentially private feature selection via stability arguments, and the robustness of the lasso. In Shalev - Shwartz, S. and Steinwart, I. (eds.), COLT - The 26th Annual Conference on Learning Theory , JMLR Workshop and Conference Proceedings. J...

    2013

  32. [40]

    Ullman, J. R. Private multiplicative weights beyond linear queries. In Milo, T. and Calvanese, D. (eds.), Proceedings of the 34th ACM Symposium on Principles of Database Systems, PODS . ACM , 2015

    2015

  33. [41]

    and Harremo \" e s, P

    van Erven, T. and Harremo \" e s, P. R \' e nyi divergence and kullback-leibler divergence. IEEE Trans. Inf. Theory , 2014

    2014

  34. [42]

    Differentially private empirical risk minimization revisited: Faster and more general

    Wang, D., Ye, M., and Xu, J. Differentially private empirical risk minimization revisited: Faster and more general. CoRR, abs/1802.05251, 2018

    2018 arXiv

  35. [43]

    H., Farokhi, F., Jin, S., Quek, T

    Wei, K., Li, J., Ding, M., Ma, C., Yang, H. H., Farokhi, F., Jin, S., Quek, T. Q. S., and Poor, H. V. Federated learning with differential privacy: Algorithms and performance analysis. IEEE Trans. Inf. Forensics Secur. , 2020

    2020

  36. [44]

    Wu, X., Li, F., Kumar, A., Chaudhuri, K., Jha, S., and Naughton, J. F. Bolt-on differential privacy for scalable stochastic gradient descent-based analytics. In Salihoglu, S., Zhou, W., Chirkova, R., Yang, J., and Suciu, D. (eds.), Proceedings of the 2017 ACM International Con...

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.