Pith. sign in

REVIEW 2 major objections 4 minor 36 references

Optimal Federated Learning for Functional Mean Estimation under Heterogeneous Privacy Constraints

T0 review · 2 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper establishes that the minimax risk of federated differentially private mean-curve estimation is set by a single fixed-point number $D_{\ast}$, giving exact rates for both shared and per-user measurement grids.

desk verdict Strong minimax theory for private functional mean estimation, but the lower-bound lemma has a real typo that must be fixed before the theorems are accepted as stated. read the letter →

arxiv 2412.18992 v2 pith:FZKWIJW7 submitted 2024-12-25 math.ST cs.LGstat.TH

Tony Cai , Abhinav Chakraborty , Lasse Vuursteen This is my paper · ORCID
classification math.STcs.LGstat.TH MSC 62G0562G2062R10
keywords FederatedlearningDifferentialprivacyFunctionalmeanestimationMinimaxratesHeterogeneousconstraintsWaveletprojectionLocalpolynomialregressionPrivacy-accuracytrade-off
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 establishes the minimax risk for estimating a population mean function from discretely sampled curves held across several servers, each with its own number of subjects, number of measurements, and differential-privacy budget. The central claim is that the risk is, up to logarithmic factors, $\eta_N D_{\ast}^{-2\alpha}$, where $D_{\ast}$ is the solution to a fixed-point equation that balances measurement density, sample size, and privacy noise across servers. The paper also shows that when each individual has private, random design points (independent design), the privacy-dominated rate is faster than when all individuals share one grid (common design). Two one-shot algorithms—a clipped wavelet projection with anisotropic Gaussian noise and a bagged local-polynomial smoother—attain these rates. If correct, this gives a sharp, server-by-server account of the privacy-accuracy trade-off in federated functional data analysis.

What carries the argument

The rate-determining object is $D_{\ast}$, the 'effective dimension' solving the fixed-point equations; $D_{\ast}$ determines the optimal truncation level $L=\log_2 D_{\ast}$ and can be read as the point where design complexity and statistical/privacy information balance. The upper bound for independent design is a truncated wavelet projection estimator: per-subject contributions are clipped at level $\tau_l^{(s)}$, rescaled coordinate-wise by $(\tau_l^{(s)}\sqrt{2^l\wedge m_s})^{-1}$, and perturbed with Gaussian noise whose variance is set by the $\ell^2$ sensitivity $c_A\sqrt{L}/n_s$, after which servers' coefficients are aggregated with inverse-variance weights $u_l^{(s)}=n_sm_s\wedge m_s2^{l(2\alpha+1)}\wedge 2^{-l}m_sn_s^2\varepsilon_s^2\wedge 2^{2l\alpha}n_s^2\varepsilon_s^2$. The common-design upper bound uses privatized per-grid-point means with Gaussian noise of variance proportional to $m/(n_s^2\varepsilon_s^2)$, followed by bagging (grouping the $m$ points into $m_0=D_{\ast}$ sized groups) and local polynomial regression. The lower bounds are built on a multivariate van Trees inequality: the minimax risk is bounded below by $2^{2L}/(\sup_f \sum_s \min\{\mathrm{Tr}(I^{Y|T}), \mathrm{Tr}(I^Y), \mathrm{Tr}(I^{X|T}), \mathrm{Tr}(I^X)\} + 2^{2L(\alpha+2)})$, and each Fisher-information trace is controlled through differential-privacy data-processing inequalities, with the random function $X_i$ constructed via $\mathrm{Beta}(2,2)$ coefficients to isolate the privacy bottleneck.

What would settle it

In the homogeneous common-design setting with fixed $n$, small $\varepsilon$, and $\alpha>1/2$, the theory predicts the privacy term dominates and the MSE should be independent of $m$; run Algorithm 2 with $n=200$, $\varepsilon=0.5$, $\alpha=1.5$, $\delta=1/n^2$, and $m=32,64,128$ on curves with Hölder regularity $1.5$ and compare MSE. If the MSE keeps decreasing as $m$ grows in that regime, the fixed-point rate is not the true limit.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the minimax integrated squared error for federated differentially private estimation of a mean function $f$ in the Hölder class $H^\alpha(R)$ is $\eta_N D_{\ast}^{-2\alpha}$, with $D_{\ast}$ defined by the fixed-point equation $D^{2\alpha}=m^{2\alpha}\wedge\sum_s \min(n_s, D^{-1}n_s^2\varepsilon_s^2)$ in the common-design case and by $D_{\ast}^{2\alpha}=\inf_{1\le D\le D_{\ast}}\sum_s \min(D^{-1}n_sm_s, D^{-2}m_sn_s^2\varepsilon_s^2, D^{2\alpha}n_s, D^{2\alpha-1}n_s^2\varepsilon_s^2)$ in the independent-design case. The four terms in the independent-design equation correspond to four information bottlenecks: sampling error, measurement error, and their privacy-constrained counterparts. In the homogeneous setting these formulas yield explicit rates, and a comparison shows that whenever privacy terms dominate, independent designs converge strictly faster than common designs, so protecting the design points does not come at a statistical cost. The paper proves matching upper and lower bounds up to polylogarithmic factors, and the lower-bound argument introduces Fisher-information data-processing inequalities that handle server heterogeneity.

Load-bearing premise

The fixed-point rate $D_{\ast}^{-2\alpha}$ is proven only for equispaced grids in the common design and i.i.d. uniform design points in the independent design; the paper only sketches the extension to general densities, so clustered or server-dependent sampling designs could break the claimed rate.

Editorial extensions

If this is right

  • The privacy-accuracy trade-off in federated functional mean estimation is fully quantified: for any heterogeneous configuration of subjects, measurements, and budgets, the minimax rate is read off from a single fixed-point equation.
  • In the common-design regime, extra measurements per subject do not improve the rate once privacy constraints bind, whereas in the independent-design regime they do; privacy-conscious studies should prefer independent measurement schedules.
  • The proposed algorithms are one-shot—each server transmits a single transcript—so they fit existing horizontal federated architectures without interactive rounds.
  • Sending $\varepsilon_s\to\infty$ recovers the classical non-private minimax rates for discretely sampled functional mean estimation, and the $S=1$ and $n_s=1$ limits recover central and local differential privacy as special cases.
  • The independent-design result provides new minimax rates even in the non-private heterogeneous case where different subjects have different numbers of measurements.

Reading between the lines

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

  • Solving the fixed-point equation in advance could be used by a system designer to allocate privacy budgets across servers: a server contributes to the rate only through its summand in the fixed-point equation, so budget should be spent where the summand is currently smallest.
  • The common-versus-independent comparison suggests a testable policy implication: adding random jitter to otherwise shared measurement grids may lower privacy cost despite making the design points sensitive information, because the diversity of design points dominates the extra noise needed to protect them.
  • The Fisher-information data-processing technique for federated differential privacy likely extends to other functional targets—covariance operators, regression functions, or derivative curves—where similar fixed-point rates with four-way bottlenecks would appear.
  • For non-uniform designs the paper only sketches an adaptation; an explicit construction with density-whitening and privacy-preserving histogram estimation would make the independent-design result fully general.
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

2 major / 4 minor

Summary. The paper studies federated functional mean estimation under heterogeneous federated differential privacy. For two designs, a common equispaced design and an independent uniform design, it states minimax rates governed by the fixed points D* defined in Eqs. (2) and (3), proposes two algorithms (a wavelet thresholding estimator for the independent design and a bagged local-polynomial estimator for the common design), and claims matching upper and lower bounds up to polylogarithmic factors. Theorems 1 and 2 give the minimax characterizations, Theorems 3 and 4 give the algorithmic upper bounds, and Theorems 5 and 6 give the lower bounds. The paper also compares the two designs in a homogeneous setting and illustrates the rates with simulations.

Significance. If the stated results hold, this is a substantial contribution: it extends federated-DP nonparametric theory to functional mean estimation, handles server-level heterogeneity in sample sizes, measurement counts, and privacy budgets, and reveals a structural advantage of independent designs over common designs in the privacy-dominated regime. A notable strength is that the rate-determining quantity D* is an implicit function of model parameters only, not fitted to data, and the homogeneous rates reduce to the known non-private rates of Cai and Yuan when the privacy constraints are removed. The paper also provides explicit, implementable algorithms and a detailed supplementary proof package. These features make the paper a useful step toward understanding the fundamental cost of privacy in functional data analysis.

major comments (2)
  1. [§4.1, Eq. (18); Supplement §8.1.1] The Van Trees prior term in Lemma 3 is displayed as 2^{2L(α+2)} in the main text, but the prior described in Supplement §8.1.1 is supported on [−2^{-L(α+1/2)}R, 2^{-L(α+1/2)}R] with per-coordinate Fisher information π²/a² = π² 2^{2L(α+1)}/R². The correct prior term is therefore of order 2^{2L(α+1)}, and as written Eq. (18) yields a lower bound of order D^{-2α-2}, not D^{-2α}. This is a load-bearing inconsistency because Lemma 3 is the stated engine for Theorems 1, 2, 5, and 6. The supplement's own algebra after Lemma 3 reduces to a denominator of order 2^{2αL}, which is consistent with the corrected exponent, suggesting the displayed value is a typo; nevertheless, the main-text lemma must be corrected and the proofs must use the corrected statement explicitly. The same issue appears in §4.2 and Supplement §8.2, where the prior term is written as π²(2L)^{2α+2}.
  2. [§2.2, Eq. (3)] The fixed point defining D* in Eq. (3) is not shown to have a unique solution. Unlike Eq. (2), for which the text gives a monotonicity argument, Eq. (3) defines D* as 'the largest number such that' an equality involving an infimum holds, but no existence or uniqueness proof is supplied. Since Theorems 2 and 5 depend on this quantity, the authors should either prove that a unique solution exists under the stated assumptions or add explicit conditions under which it does.
minor comments (4)
  1. [§5.3] The text 'Similarly to the dependent design case' should read 'Similarly to the independent design case'; this is a typographical error in the caption discussion.
  2. [Algorithms 1 and 2] The notation for the privacy parameter in the Gaussian mechanism is inconsistent: Algorithm 1 uses δ in the noise variance while Algorithm 2 uses δ_s. Since the servers are heterogeneous, δ_s should be used in both places.
  3. [§1.1] The claim that the uniform-design assumption can be relaxed to non-uniform distributions 'bounded above and below by a constant' is only asserted, not demonstrated. The stated theorems are for uniform or equispaced designs; please either add the non-uniform derivation or explicitly state the restricted scope of the theorems.
  4. [Supplement §8.1.1] In the proof of Lemma 3, the sentence 'Combining this with the earlier identities of Lemmas 3 and 4' refers to Lemma 3 itself; this should be rephrased to refer to the relevant trace bounds and the main-text Lemma 4.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: D* is a model-parameter fixed point, upper and lower bounds are proved independently, and the self-citations to [8] supply technical tools rather than the assumed conclusion.

full rationale

Walking the derivation chain, the central rates are not assumed. D* in (2)/(3) is defined as the solution of a fixed-point equation in the model parameters (m, ns, εs, α); it is not fitted from data and is not the target of estimation. Theorems 3 and 4 prove risk bounds after optimizing over resolution or group size: the upper-bound calculations produce expressions of the form Σ_s min(...) + 2^{-2Lα}, and choosing L = ⌈log₂ D*⌉ yields (D*)^{-2α}; the lower bounds (Lemma 3 and Theorems 5–6) independently derive matching lower bounds via the multivariate van Trees inequality [17] and Fisher-information bounds, so the equality of upper and lower rates is a proved minimax result, not an identity by construction. The non-private limit ε_s → ∞ recovers the externally established Cai–Yuan [9] rates, providing a benchmark outside the present paper. Self-citations to [8] are present: Definition 1 adopts the FDP framework from [8], and the proof of Lemma 15 invokes 'Lemma 4.2 in [8]' for a Fisher-information data-processing bound. These are borrowed tools from prior work of overlapping authorship, and the cited lemma is not the functional-mean minimax conclusion; the paper's own supplement supplies substantial parts of the Fisher-information analysis (e.g., Lemmas 13–17). Thus the self-citations are not load-bearing in the sense of assuming the target theorem. A separate correctness concern should be recorded: Lemma 3's displayed denominator 2^{2L(α+2)} is inconsistent with the supplement's prior Fisher information (order 2^{2L(α+1)}), and if taken literally would change the rate; this is a proof error, not a circularity, and does not raise the circularity score.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The central claims rest on Hölder smoothness, specific design distributions, Gaussian noise, and a small-δ regime; all are stated assumptions rather than fitted parameters. The proofs additionally invoke standard wavelet and Van Trees machinery. No ad hoc parameters are fitted to data: the fixed-point D* and the algorithm weights are deterministic functions of the model parameters (n_s, m_s, ε_s, α). No new latent entities are introduced.

assumptions (7)
  • domain assumption The random functions X_i belong to a Hölder ball H^α(R) on [0,1] with α > 1/2.
    Section 1.1 defines the function class; the rates D^{-2α} and the wavelet coefficient decay all rely on this smoothness.
  • domain assumption Independent design points are i.i.d. Uniform[0,1]; common design is equispaced for the lower bound.
    Sections 1.1 and 2.1. The minimax rates are proved for these sampling schemes; relaxation to bounded-density designs is only sketched.
  • domain assumption Measurement noise is Gaussian: ξ_ij ~ N(0,1).
    Section 1.1 sets σ=1 without loss of generality; Gaussianity is used in Fisher information and likelihood calculations in Sections 4 and 8.
  • domain assumption The small-δ condition δ' log(1/δ') ≲ (n_s/m_s ∧ sqrt(n_s/m_s)) ε_s^2 / N holds.
    Stated in Theorems 1, 2, 5, 6 and used in Lemmas 15-17 to control tail terms in the lower-bound data processing inequalities.
  • standard math An A>α-regular compactly supported orthonormal wavelet basis is available (Daubechies wavelets).
    Section 3.1 uses this basis to construct the projection estimators and to characterize H^α; existence is classical.
  • domain assumption Assumptions (LP1) and (LP2) on the local design points within each bagging group hold.
    Section 3.2 and Theorem 4; these regularity conditions are needed for the local polynomial weights W* and the bias/variance bounds.
  • standard math The multivariate Van Trees inequality (Gill-Levit 1995) and the Fisher information data processing inequalities (Zamir 1998) apply.
    Section 4 and Supplementary Section 8 use these cited results for the lower bounds; they are not proved in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimal Federated Learning for Functional Mean Estimation under Heterogeneous Privacy Constraints." pith.science (2026). https://pith.science/paper/FZKWIJW7

@misc{pith2026241218992,
  author       = {Pith},
  title        = {Pith review of: Optimal Federated Learning for Functional Mean Estimation under Heterogeneous Privacy Constraints},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FZKWIJW7}},
  note         = {Machine review of arXiv:2412.18992}
}
read the original abstract

Federated learning (FL) is a distributed machine learning technique designed to preserve data privacy and security, and it has gained significant importance due to its broad range of applications. This paper addresses the problem of optimal functional mean estimation from discretely sampled data in a federated setting. We consider a heterogeneous framework where the number of individuals, measurements per individual, and privacy parameters vary across one or more servers, under both common and independent design settings. In the common design setting, the same design points are measured for each individual, whereas in the independent design, each individual has their own random collection of design points. Within this framework, we establish minimax upper and lower bounds for the estimation error of the underlying mean function, highlighting the nuanced differences between common and independent designs under distributed privacy constraints. We propose algorithms that achieve the optimal trade-off between privacy and accuracy and provide optimality results that quantify the fundamental limits of private functional mean estimation across diverse distributed settings. These results characterize the cost of privacy and offer practical insights into the potential for privacy-preserving statistical analysis in federated environments.

Figures

Figures reproduced from arXiv: 2412.18992 by the authors.

Figure 1
Figure 1. MSE vs n (with fixed m = 64) and MSE vs m (with fixed n = 200) for Independent design The intuitively obvious finding that can be observed from both panels in [PITH_FULL_IMAGE:figures/full_fig_p022_1.png] view at source ↗
Figure 2
Figure 2. Log-scale MSE vs n (with fixed m = 64) and vs m (with fixed n = 200) for the common design setting. Another stark difference compared to the independent design case, is that the estimation error plateaus much quicker as m increases, transitioning to a regime where the rate becomes independent of m. This is in line with the main results (see Corollary 2), which state that there is only a benefit of having more measur… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 32 canonical work pages

  1. [33]

    Optimal estimation in private distributed functional data analysis

    Gengyu Xue, Zhenhua Lin, and Yi Yu. Optimal estimation in private distributed functional data analysis. 2024

    work page 2024

  2. [1]

    Op- timal rates for nonparametric density estimation under communication constraints

    Jayadev Acharya, Cl´ ement L Canonne, Aditya Vikram Singh, and Himanshu Tyagi. Op- timal rates for nonparametric density estimation under communication constraints. IEEE Transactions on Information Theory , 2023

    work page 2023

  3. [2]

    Unified lower bounds for interactive high-dimensional estimation under information constraints.Advances in Neural Information Processing Systems , 36, 2024

    Jayadev Acharya, Cl´ ement L Canonne, Ziteng Sun, and Himanshu Tyagi. Unified lower bounds for interactive high-dimensional estimation under information constraints.Advances in Neural Information Processing Systems , 36, 2024

    work page 2024

  4. [3]

    Discrete distribution estimation under user-level local differential privacy

    Jayadev Acharya, Yuhan Liu, and Ziteng Sun. Discrete distribution estimation under user-level local differential privacy. In Proceedings of The 26th International Conference on Artificial Intelligence and Statistics , volume 206, pages 8561–8585. PMLR, 25–27 Apr 2023

    work page 2023

  5. [4]

    Differentially private assouad, fano, and le cam

    Jayadev Acharya, Ziteng Sun, and Huanyu Zhang. Differentially private assouad, fano, and le cam. In Algorithmic Learning Theory, pages 48–78. PMLR, 2021. 23

    work page 2021

  6. [5]

    Fisher information under local differential privacy

    Leighton Pate Barnes, Wei-Ning Chen, and Ayfer ¨Ozg¨ ur. Fisher information under local differential privacy. IEEE Journal on Selected Areas in Information Theory , 1(3):645–659, 2020

    work page 2020

  7. [6]

    Federated learning for emoji prediction in a mobile keyboard

    Francoise Beaufays, Kanishka Rao, Rajiv Mathews, and Swaroop Ramaswamy. Federated learning for emoji prediction in a mobile keyboard. arXiv:1906.04329, 2019

    arXiv 1906

  8. [7]

    Tony Cai, Abhinav Chakraborty, and Lasse Vuursteen

    T. Tony Cai, Abhinav Chakraborty, and Lasse Vuursteen. Federated nonparametric hy- pothesis testing with differential privacy constraints: Optimal rates and adaptive tests. 2024

    work page 2024

Show all 36 references
  1. [8]

    Optimal federated learning for nonparametric regression with heterogeneous distributed differential privacy constraints

    T Tony Cai, Abhinav Chakraborty, and Lasse Vuursteen. Optimal federated learning for nonparametric regression with heterogeneous distributed differential privacy constraints. arXiv preprint arXiv:2406.06755 , 2024

    2024 arXiv

  2. [9]

    Tony Cai and Ming Yuan

    T. Tony Cai and Ming Yuan. Optimal estimation of the mean function based on discretely sampled functional data: Phase transition. The Annals of Statistics , 39(5):2330 – 2355, 2011

    2011

  3. [10]

    Private distribution testing with heterogeneous constraints: Your epsilon might not be mine

    Cl´ ement L Canonne and Yucheng Sun. Private distribution testing with heterogeneous constraints: Your epsilon might not be mine. In 15th Innovations in Theoretical Computer Science Conference (ITCS 2024) , 2024

    2024

  4. [11]

    Jordan, Xuelin Yang, Lydia Zakynthinou, and Nikita Zhivotovskiy

    Yuval Dagan, Michael I. Jordan, Xuelin Yang, Lydia Zakynthinou, and Nikita Zhivotovskiy. Dimension-free private mean estimation for anisotropic distributions. 2024

    2024

  5. [12]

    Ten lectures on wavelets

    Ingrid Daubechies. Ten lectures on wavelets . SIAM, 1992

    1992

  6. [13]

    Constructive Approximation

    Ronald A DeVore and George G Lorentz. Constructive Approximation. Springer, 1993

    1993

  7. [14]

    Calibrating noise to sensitivity in private data analysis

    Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith. Calibrating noise to sensitivity in private data analysis. In Theory of Cryptography: Third Theory of Cryptog- raphy Conference, TCC 2006, New York, NY, USA, March 4-7, 2006. Proceedings 3 , pages 265–284. Springer, 2006

    2006

  8. [15]

    The algorithmic foundations of differential privacy

    Cynthia Dwork, Aaron Roth, et al. The algorithmic foundations of differential privacy. Foundations and Trends® in Theoretical Computer Science , 9(3–4):211–407, 2014

    2014

  9. [16]

    Springer Science & Business Media, Berlin, Heidelberg, 2006

    Fr´ ed´ eric Ferraty and Philippe Vieu.Nonparametric Functional Data Analysis: Theory and Practice. Springer Science & Business Media, Berlin, Heidelberg, 2006

    2006

  10. [17]

    Applications of the van trees inequality: a bayesian cram´ er-rao bound.Bernoulli, pages 59–79, 1995

    Richard D Gill and Boris Y Levit. Applications of the van trees inequality: a bayesian cram´ er-rao bound.Bernoulli, pages 59–79, 1995

    1995

  11. [18]

    Differential privacy for functions and functional data

    Rob Hall, Alessandro Rinaldo, and Larry Wasserman. Differential privacy for functions and functional data. The Journal of Machine Learning Research , 14(1):703–727, 2013. 24

    2013

  12. [19]

    Wavelets, approximation, and statistical applications, volume 129

    Wolfgang H¨ ardle, Gerard Kerkyacharian, Dominique Picard, and Alexander Tsybakov. Wavelets, approximation, and statistical applications, volume 129. Springer Science & Busi- ness Media, 2012

    2012

  13. [20]

    Sampling depth trade-off in function esti- mation under a two-level design

    Akira Horiguchi, Li Ma, and Botond T Szab´ o. Sampling depth trade-off in function esti- mation under a two-level design. arXiv preprint arXiv:2310.02968 , 2023

    2023 arXiv

  14. [21]

    Function Estimation and Gaussian Sequence Models

    Iain M Johnstone. Function Estimation and Gaussian Sequence Models . Unpublished manuscript, 2019

    2019

  15. [22]

    On the statistical complexity of estimation and testing under privacy constraints

    Cl´ ement Lalanne, Aur´ elien Garivier, and R´ emi Gribonval. On the statistical complexity of estimation and testing under privacy constraints. Transactions on Machine Learning Research Journal, 2023

    2023

  16. [23]

    Learning with user-level privacy.Advances in Neural Information Processing Systems, 34:12466–12479, 2021

    Daniel Levy, Ziteng Sun, Kareem Amin, Satyen Kale, Alex Kulesza, Mehryar Mohri, and Ananda Theertha Suresh. Learning with user-level privacy.Advances in Neural Information Processing Systems, 34:12466–12479, 2021

    2021

  17. [24]

    Learning discrete distributions: user vs item-level privacy

    Yuhan Liu, Ananda Theertha Suresh, Felix Xinnan X Yu, Sanjiv Kumar, and Michael Riley. Learning discrete distributions: user vs item-level privacy. Advances in Neural Information Processing Systems, 33:20965–20976, 2020

    2020

  18. [25]

    Formal privacy for func- tional data with gaussian perturbations

    Ardalan Mirshani, Matthew Reimherr, and Aleksandra Slavkovi´ c. Formal privacy for func- tional data with gaussian perturbations. In International Conference on Machine Learning, pages 4595–4604. PMLR, 2019

    2019

  19. [26]

    Tight and robust private mean estimation with few users

    Shyam Narayanan, Vahab Mirrokni, and Hossein Esfandiari. Tight and robust private mean estimation with few users. In International Conference on Machine Learning , pages 16383–16412. PMLR, 2022

    2022

  20. [27]

    Principal components analysis for functional data

    JO Ramsay and BW Silverman. Principal components analysis for functional data. Func- tional data analysis , pages 147–172, 2005

    2005

  21. [28]

    Estimating the mean and covariance structure nonparametrically when the data are curves

    John A Rice and Bernard W Silverman. Estimating the mean and covariance structure nonparametrically when the data are curves. Journal of the Royal Statistical Society: Series B (Methodological), 53(1):233–243, 1991

    1991

  22. [29]

    Tsybakov

    A. Tsybakov. Introduction to nonparametric estimation. In Springer Series in Statistics , 2008

    2008

  23. [30]

    Tsybakov

    Alexandre B. Tsybakov. Introduction to nonparametric estimation. Springer series in statis- tics. Springer, New York ; London, 2009. OCLC: ocn300399286

    2009

  24. [31]

    High-Dimensional Probability: An Introduction with Applications in Data Science

    Roman Vershynin. High-Dimensional Probability: An Introduction with Applications in Data Science. Cambridge University Press, 1 edition, September 2018. 25

    2018

  25. [32]

    Functional data analysis

    Jane-Ling Wang, Jeng-Min Chiou, and Hans-Georg M¨ uller. Functional data analysis. An- nual Review of Statistics and its application , 3(1):257–295, 2016

    2016

  26. [34]

    ∇fL log dPT f dµ′ !#

    R. Zamir. A proof of the fisher information inequality via a data processing argument. IEEE Transactions on Information Theory , 44(3):1246–1250, 1998. Supplementary Material 6 Proofs related to the upper bound theorems 6.1 Proof for the independent design upper bound We first...

    1998

  27. [35]

    Lemma 14

    The result now follows: Φ (Σ +Im)−1 ΦT ⪯ ∥ϕd∥2 2 − c2 L∥ϕd∥4 2 1 + c2 L∥ϕd∥2 2 I2L = ∥ϕd∥2 2 1 + c2 L∥ϕd∥2 2 I2L ⪯ 1 c2 L I2L ⪯ 2L∥ϕ∥2 ∞I2L. Lemma 14. For M ≍ 2L(√msns ∨ ms) log(1/δs) we have that 2M δs + Z ∞ M P (|G1| ≥t) dt + Z ∞ M P | ˘G1| ≥t dt ≲ δs2L(√msns ∨ ms) log(1/δs)...

  28. [36]

    exp t SY (s) 1 f 2 2 ! exp 1 2 t2 c2 ϕ ms (ns − 1) 2L SY (s) 1 f 2 2 !# = 2 Ef

    Using the linearity of expectation, we have that ⟨SY (s) f , SY (s) 1 f ⟩ = Pns i=1⟨Sf (Zi), SY (s) 1 f ⟩. We can split the sum into two terms ⟨SY (s) f , SY (s) 1 f ⟩ = ∥SY (s) 1 f ∥2 2 + nsX i=2 msX j=1 ξij⟨Sf (Zi), SY (s) 1 f ⟩ Hence we have that Eet|⟨SY (s) f ,S Y (s) i f ...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.