Pith. sign in

REVIEW 2 major objections 3 minor 300 references

Stochastic nonconvex optimization can be reduced in a black-box way to static-regret online learning, achieving SGD-like rates and the optimal nonsmooth rate.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 17:32 UTC pith:GSGJNFPC

load-bearing objection The smooth black-box reduction is a genuine contribution and likely resolves the Chen-Hazan open problem on that side, but the nonsmooth O(T^{-2/7}) claim is proved only for a weaker mean-gradient stationarity measure, so the headline as stated does not follow. the 2 major comments →

arxiv 2607.17607 v1 pith:GSGJNFPC submitted 2026-07-20 cs.LG math.OCstat.ML

Optimizing the Preconditioner: A Black-box Online-to-Nonconvex Conversion with Static Regret Minimization Oracles

Haichen Hu , David Simchi-Levi This is my paper
classification cs.LG math.OCstat.ML MSC 90C2690C1568W27
keywords nonconvex optimizationonline convex optimizationstatic regretonline-to-nonconvex conversionpreconditioningGoldstein stationaritystochastic gradient oracleadaptive methods
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper tries to establish that offline stochastic nonconvex optimization can be reduced, as a black box, to ordinary static-regret minimization in online convex optimization. The key idea is to let the online learner choose a preconditioner—not the update direction—while a predictable gradient tracker supplies the descent signal. For smooth objectives, any online algorithm with O(√T) static regret against the identity yields the classical O(1/√T) convergence rate for finding approximate stationary points. For nonsmooth Lipschitz objectives, the same framework yields O(T^(-2/7)) for a Goldstein-type stationarity measure, which the paper argues is optimal. If correct, the results resolve a 2024 open problem about whether adaptive gradient methods can be understood through static regret alone.

Core claim

Central claim: online-to-nonconvex conversion can run on undiscounted static regret against a single fixed comparator, not dynamic or shifting regret. The algorithm keeps a moving-average gradient tracker m_t, draws a fresh stochastic gradient s_t, and feeds the linear loss L_t(U) = -⟨s_t, U m_t⟩/G to an online learner choosing a preconditioner U_t; the iterate updates via x_{t+1} = x_t - η U_t m_t. Unbiasedness and linearity let smoothness turn any static-regret bound against identity into a bound on the average squared gradient norm; the regret term measures the cost of learning the preconditioner versus plain SGD. The nonsmooth extension adds a random segment query and a normalized direct

What carries the argument

The load-bearing mechanism is the preconditioner-selection interface: a compact convex set U of symmetric matrices with 0 ⪯ U ⪯ I and I ∈ U. The online learner plays U_t and pays linear losses L_t(U) = -⟨s_t, U m_t⟩/G (smooth case) or L_t(U) = -⟨s_t, U z_t⟩ (nonsmooth case); static regret against the identity comparator measures how much an adaptive preconditioner beats plain gradient descent. The tracker m_t, an exponential moving average of gradients projected to the natural gradient scale, converts regret into stationarity. For nonsmooth objectives, a smoothly normalized direction z_t and a random segment query replace the unavailable Lipschitz-gradient descent lemma.

Load-bearing premise

The nonsmooth O(T^{-2/7}) result is proven only for a relaxed stationarity measure that permits opposing gradient directions to cancel inside the expectation, while the main text states the stronger Goldstein measure with the norm inside the expectation; if the intended measure is the standard one, the rate does not follow from the provided proof.

What would settle it

Construct a 1-D Lipschitz function whose gradient alternates between +1 and -1 on an interval of length comparable to the algorithm's displacement. At the output point, the appendix measure's E[∇f] can be zero while the main-text measure's E|∇f| is one, so the bound (C.52) would certify stationarity under the relaxed measure but not under Definition 7.2. Checking this gap numerically or analytically would settle whether the claimed optimal rate holds for the stated notion.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Any OCO algorithm with O(√T) static regret on the induced linear losses immediately yields a stochastic nonconvex optimizer with the classical O(1/√T) stationarity rate, without further structural assumptions on the algorithm.
  • Adaptive methods like diagonal AdaGrad and Shampoo become instances of the reduction; their data-dependent regret bounds translate directly into data-dependent nonconvex convergence guarantees.
  • The nonsmooth result indicates that the optimal O(T^{-2/7}) rate for Goldstein stationarity is achievable purely from static regret, matching known lower bounds.
  • The separation of gradient prediction from preconditioner selection offers a modular design principle: changing the online learner changes the geometry without altering the descent logic.
  • A negative static regret against the identity means the learned preconditioner outperforms vanilla gradient descent on the induced losses, giving a measurable notion of adaptation benefit.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The nonsmooth proof's stationarity measure (appendix eq. C.1: inf over distributions with mean x of ||E∇f(X)|| + λE||X-x||²) is weaker than the main text's Definition 7.2 (E||∇f(w)|| + λE||x-w||²). Under the stated definition, the O(T^{-2/7}) guarantee may not follow from the appended argument, since opposing gradient directions cancel in the appendix measure. This discrepancy is editorially infer
  • The black-box guarantee for 'any' static-regret oracle is only as good as the oracle's regret on the specific adaptive linear losses generated by the reduction; the paper's own instantiations are data-adaptive algorithms (AdaGrad, Shampoo), so an arbitrary low-regret oracle without second-moment control is not automatically covered.
  • The linear-loss interface could serve as a template for designing new optimizers by choosing different preconditioner sets U (e.g., group-structured or low-rank), a direction the paper gestures at but does not pursue.
  • A direct empirical test would be to run Algorithm 1 with a simple static-regret online learner and compare the average squared gradient norm against SGD on nonconvex ML training; if the regret term dominates, the practical benefit of the preconditioner slot would be measurable.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 3 minor

Summary. The paper proposes a black-box reduction from stochastic nonconvex optimization to static regret minimization in online convex optimization. Algorithm 1 maintains a predictable gradient tracker m_t and uses an arbitrary OCO oracle over a compact matrix preconditioner class to choose U_t, with update x_{t+1}=x_t−η U_t m_t. Theorem 6.1 bounds the average squared gradient norm by O(σ√(Mβ)/√T)+O(√(Mβ) R_T(A,I_d)/T + Mβ/T), so an oracle with R_T=O(√T) recovers the classical O(1/√T) rate. Section 7 extends the framework to Lipschitz nonsmooth objectives via random-segment queries and a weighted output, claiming an optimal O(T^{-2/7}) rate for a Goldstein-type stationarity measure. Section 8 instantiates diagonal AdaGrad and projected Shampoo as concrete oracles. The smooth-part proof is in Appendix B; the nonsmooth proof is in Appendix C.

Significance. The smooth reduction is a conceptually valuable contribution: it cleanly separates gradient tracking from preconditioner selection, uses only undiscounted static regret, and is proved with explicit constants. If the nonsmooth theorem were supported for the stated stationarity measure, the paper would resolve the Chen–Hazan open problem in both smooth and nonsmooth settings, which is a substantial advance. The concrete AdaGrad/Shampoo regret transfers are useful. However, the nonsmooth result is proved only for a strictly weaker mean-gradient measure than the one defined in the main text, so the paper's headline optimal-rate claim for Goldstein stationarity is not currently established.

major comments (2)
  1. [Definition 7.2 vs. Appendix C, eq. (C.1); Lemma C.7, eq. (C.52)] Definition 7.2 states the Goldstein-type measure with E[||∇f(w)||] (norm inside the expectation), while Appendix C, eq. (C.1), defines the weaker measure with ||E[∇f(X)]||. Lemma C.7's inequality (C.52) bounds stationarity by ||Σ_s q_{t,s} g_s|| + λΣ...; this is valid only for (C.1). Under Definition 7.2, the first term must be Σ_s q_{t,s}||g_s||, and the proof controls only the smaller quantity ||Σ_s q_{t,s}g_s||. Since ||Σqg|| ≤ Σq||g||, opposing gradients can cancel; no bound on the Goldstein measure follows. Thus Theorem 7.3's O(T^{-2/7}) claim is unproven for the stated measure.
  2. [Abstract and Section 9] The optimal-rate claim is explicitly tied to a 'Goldstein stationary point' and to resolving Chen–Hazan's open problem in the nonsmooth setting. Because the proof only supports the relaxed mean-gradient measure (C.1), these statements overclaim. If the relaxed measure is the intended contribution, Definition 7.2 and Theorem 7.3 must be revised and the optimality claim re-scoped. This is not a presentation issue: it changes what is proved.
minor comments (3)
  1. [Lemma B.3 proof] The line '∥e_1∥² ≤ G_0² = 2(Mβ)²' should read '2Mβ'; the subsequent algebra uses 2G² = 2Mβ, so this is a typographical error, but it should be corrected.
  2. [Algorithm 2, output distribution] The verification that the output probabilities sum to one appears only in Lemma C.5; a one-line statement in the main text would improve readability.
  3. [Section 8.1] The data-dependent AdaGrad regret bound is derived under ∥m_t∥₂ ≤ √2 G. This is correct via Lemma B.1, but it should be stated explicitly before the inequality, since it is used to pass from q_{t,i} to s_{t,i}.

Circularity Check

0 steps flagged

No circularity: the main theorem is a genuine reduction whose regret term is an explicit interface assumption, with no fitted parameters, self-citation loops, or definitional equivalences.

full rationale

The paper's central result, Theorem 6.1, derives a bound on average squared gradient norm that decomposes into a stochastic-noise term and the static regret R_T(A, I_d) of a black-box OCO oracle on the algorithm's own linear losses. This is not circular: R_T is an input assumption about the oracle, not a quantity derived from the stationarity measure, and the bound is a genuine implication of smoothness, tracking, and the regret inequality. The claim that any OCO oracle with R_T = O(sqrt(T)) recovers O(1/sqrt(T)) follows by algebra from the stated bound, not by construction. Similarly, the nonsmooth Theorem 7.3 reduces stationarity to a regret term plus explicit parameter-dependent terms; neither term is fitted to the target. The AdaGrad and Shampoo instantiations rely on external, independently established regret theorems (Duchi et al., 2011; Gupta et al., 2018), and the paper's own trajectory-dependent bounds are derived rather than assumed. There are no self-citations used as load-bearing evidence, no uniqueness theorems imported from the authors' prior work, and no ansatz smuggled in by citation. The reader's concern about a mismatch between Definition 7.2 and the stationarity measure in Appendix C.1 is a potential proof-correctness issue, not a circularity: Lemma C.7 may establish the result only for the mean-gradient measure ||E[∇f(X)]|| instead of E[||∇f(X)||], but this is a question of whether the appended proof matches the stated definition, not of the derivation reducing to its own inputs. Accordingly, the circularity score is 0.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

No numbers are fitted to data; a, η, ρ, ε_T are analytic functions of the stated problem constants (M, β, σ, L, λ, T). No new particles, forces, or entities are introduced. The R_T(A,I_d) term is an explicit interface assumption, not a hidden fitted parameter. The one ad hoc element is the unstated switch to the weaker stationarity measure in the nonsmooth proof.

axioms (5)
  • domain assumption f is β-smooth and has range bounded by M: f(x) − f(y) ≤ M for all x,y (Assumption 4.1).
    Drives the gradient bound ||∇f|| ≤ √(2Mβ) (Lemma B.1), the tracker radius G₀, the descent lemma, and the G²/T term. If the range is unbounded, the progress budget argument collapses.
  • domain assumption Stochastic oracle O(x) with E[∇̃f(x)|x] = ∇f(x) and E||∇̃f(x) − ∇f(x)||² ≤ σ² (Assumption 4.2).
    The noise model for every expectation and variance computation; the a = G/max{G,σ√T} scale and the σ/√T terms follow from it.
  • domain assumption The OCO oracle A has a fixed static-regret guarantee R_T(A,u) = E[Σℓ_t(u_t) − Σℓ_t(u)] against any comparator (Definition 4.3).
    The entire reduction is a wrapper around this guarantee; the final bound is linear in R_T(A,I_d). The oracle must achieve this on the constructed adaptive linear losses for the 'any black-box OCO algorithm' claim to hold.
  • domain assumption Nonsmooth setting: ||∇f|| ≤ L, range ≤ M, and the Newton-Leibniz formula holds along segments (Assumption 7.1).
    Replaces the smooth descent lemma with the exact random-segment identity in Lemma C.1; the L and M constants enter every bound in Theorem 7.3.
  • ad hoc to paper The stationarity measure used in the proof is Appendix (C.1), inf over p with mean x of ||E[∇f(X)]||₂ + λE||X−x||₂², not Definition 7.2's E[||∇f(w)||] + λE||x−w||₂².
    Lemma C.7's inequality (C.52) requires the norm-of-expectation form; the text does not flag the switch. This is the load-bearing unstated modeling choice for the nonsmooth theorem.

pith-pipeline@v1.3.0-alltime-deepseek · 29820 in / 38399 out tokens · 329148 ms · 2026-08-01T17:32:52.009325+00:00 · methodology

0 comments
read the original abstract

We study whether stochastic nonconvex optimization can be reduced to ordinary static regret minimization in online convex optimization in a black-box manner. For smooth nonconvex objectives, our reduction maintains a predictable gradient tracker, while a black-box online learner selects a preconditioner that determines how this tracker is transformed into the update direction. The learner receives linear convex losses and is evaluated against a single fixed comparator over one undiscounted online game. For a $\beta$-smooth objective with range bounded by $M$ and an unbiased stochastic-gradient oracle with variance bounded by \(\sigma^2\), we establish $$\frac{1}{T}\sum_{t=1}^T \mathbb E\!\left[\|\nabla f(x_t)\|_2^2\right] \lesssim \frac{\sigma\sqrt{M\beta}}{\sqrt T} + \frac{\sqrt{M\beta}\, \mathscr R_T(\mathcal A,I_d)}{T} + \frac{M\beta}{T}.$$ Consequently, any black-box OCO algorithm with $\mathscr R_T(\mathcal A,I_d)=O(\sqrt T)$ recovers the classical $O(\frac{1}{\sqrt{T}})$ convergence rate. We further show that the same black-box framework extends beyond the smooth setting to Lipschitz nonconvex objectives without Lipschitz continuous gradients. Importantly, this extension continues to rely only on an ordinary static-regret guarantee and requires no stronger notion of online regret. When the OCO oracle admits square-root static regret, the resulting conversion achieves the optimal $O(T^{-2/7})$ convergence rate for the corresponding Goldstein stationary point. These results resolve the open problem posed by Chen and Hazan (2024). More broadly, our framework separates optimizer design into gradient prediction and online preconditioner selection, providing a principled perspective on how adaptive optimization methods such as AdaGrad and Shampoo may be understood through static regret and applied in nonconvex optimization.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

300 extracted references · 1 canonical work pages

  1. [1]

    Mathematics of Operations Research , year=

    Bypassing the Monster: A Faster and Simpler Optimal Algorithm for Contextual Bandits under Realizability , author=. Mathematics of Operations Research , year=

  2. [2]

    2012 , publisher=

    Traces and determinants of linear operators , author=. 2012 , publisher=

    2012

  3. [3]

    WDS , volume=

    Data splitting , author=. WDS , volume=. 2010 , organization=

    2010

  4. [4]

    BMC medical genomics , volume=

    Optimally splitting cases for training and testing high dimensional classifiers , author=. BMC medical genomics , volume=. 2011 , publisher=

    2011

  5. [5]

    Cross Validation in Encyclopedia of Database Systems (ed

    Refaeilzadeh, P and Tang, L and Liu, H , year=. Cross Validation in Encyclopedia of Database Systems (ed. Liu, L &

  6. [6]

    arXiv preprint arXiv:2401.16407 , year=

    Is K-fold cross validation the best model selection method for Machine Learning? , author=. arXiv preprint arXiv:2401.16407 , year=

    Pith/arXiv arXiv

  7. [7]

    , author=

    Cross-validation. , author=

  8. [8]

    arXiv preprint arXiv:2303.08774 , year=

    Gpt-4 technical report , author=. arXiv preprint arXiv:2303.08774 , year=

    Pith/arXiv arXiv

  9. [9]

    arXiv preprint arXiv:2405.17796 , year=

    Offline Oracle-Efficient Learning for Contextual MDPs via Layerwise Exploration-Exploitation Tradeoff , author=. arXiv preprint arXiv:2405.17796 , year=

    Pith/arXiv arXiv

  10. [10]

    arXiv preprint arXiv:2107.10119 , year=

    Truthful information elicitation from hybrid crowds , author=. arXiv preprint arXiv:2107.10119 , year=

    Pith/arXiv arXiv

  11. [11]

    arXiv preprint arXiv:2402.06062 , year=

    Peer Expectation in Robust Forecast Aggregation: The Possibility/Impossibility , author=. arXiv preprint arXiv:2402.06062 , year=

    Pith/arXiv arXiv

  12. [12]

    Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics , pages=

    Contextual bandits with linear payoff functions , author=. Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics , pages=. 2011 , organization=

    2011

  13. [13]

    Journal of the American Statistical Association , volume=

    Fast calibrated additive quantile regression , author=. Journal of the American Statistical Association , volume=. 2021 , publisher=

    2021

  14. [14]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Risk-Aware Continuous Control with Neural Contextual Bandits , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  15. [15]

    2016 IEEE Congress on Evolutionary Computation (CEC) , pages=

    Contextual bandit algorithm for risk-aware recommender systems , author=. 2016 IEEE Congress on Evolutionary Computation (CEC) , pages=. 2016 , organization=

    2016

  16. [16]

    arXiv preprint arXiv:1610.05129 , year=

    Risk-aware algorithms for adversarial contextual bandits , author=. arXiv preprint arXiv:1610.05129 , year=

    Pith/arXiv arXiv

  17. [17]

    IEEE Transactions on information theory , volume=

    Sequential hypothesis test with online usage-constrained sensor selection , author=. IEEE Transactions on information theory , volume=. 2019 , publisher=

    2019

  18. [18]

    Proceedings of the 7th ACM SIGCOMM conference on Internet measurement , pages=

    Passive online rogue access point detection using sequential hypothesis testing with TCP ACK-pairs , author=. Proceedings of the 7th ACM SIGCOMM conference on Internet measurement , pages=

  19. [19]

    1997 , publisher=

    Gaussian regression and optimal finite dimensional linear models , author=. 1997 , publisher=

    1997

  20. [20]

    Trace Ideals with Applications (Lecture Notes) , author=

  21. [21]

    2005 , publisher=

    Trace ideals and their applications , author=. 2005 , publisher=

    2005

  22. [22]

    arXiv preprint arXiv:2112.13487 , year=

    The statistical complexity of interactive decision making , author=. arXiv preprint arXiv:2112.13487 , year=

    Pith/arXiv arXiv

  23. [23]

    Advances in neural information processing systems , volume=

    Bellman eluder dimension: New rich classes of rl problems, and sample-efficient algorithms , author=. Advances in neural information processing systems , volume=

  24. [24]

    arXiv preprint arXiv:2210.10278 , year=

    A reinforcement learning approach in multi-phase second-price auction design , author=. arXiv preprint arXiv:2210.10278 , year=

    arXiv

  25. [25]

    On concentration of self-bounding functions , author=

  26. [26]

    2014 , publisher=

    Mathematical methods in quantum mechanics , author=. 2014 , publisher=

    2014

  27. [27]

    , author=

    On general minimax theorems. , author=

  28. [28]

    Annali di Matematica Pura ed Applicata , volume=

    Eigenvalue decay rates for positive integral operators , author=. Annali di Matematica Pura ed Applicata , volume=. 2013 , publisher=

    2013

  29. [29]

    1978 , publisher=

    IV: Analysis of Operators , author=. 1978 , publisher=

    1978

  30. [30]

    International Conference on Artificial Intelligence and Statistics , pages=

    Sample complexity of kernel-based q-learning , author=. International Conference on Artificial Intelligence and Statistics , pages=. 2023 , organization=

    2023

  31. [31]

    arXiv preprint arXiv:2312.16730 , year=

    Foundations of reinforcement learning and interactive decision making , author=. arXiv preprint arXiv:2312.16730 , year=

    Pith/arXiv arXiv

  32. [32]

    arXiv preprint arXiv:2211.08875 , year=

    Learning linear operators: Infinite-dimensional regression as a well-behaved non-compact inverse problem , author=. arXiv preprint arXiv:2211.08875 , year=

    Pith/arXiv arXiv

  33. [33]

    arXiv preprint arXiv:2402.15715 , year=

    Operator learning: Algorithms and analysis , author=. arXiv preprint arXiv:2402.15715 , year=

    Pith/arXiv arXiv

  34. [34]

    arXiv preprint arXiv:2404.03761 , year=

    Learning smooth functions in high dimensions: from sparse polynomials to deep neural networks , author=. arXiv preprint arXiv:2404.03761 , year=

    Pith/arXiv arXiv

  35. [35]

    International Conference on Machine Learning , pages=

    Sparse learning of dynamical systems in RKHS: An operator-theoretic approach , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  36. [36]

    Advances in Neural Information Processing Systems , volume=

    Kernelized reinforcement learning with order optimal regret bounds , author=. Advances in Neural Information Processing Systems , volume=

  37. [37]

    Applied Mathematics and Computation , volume=

    Computation of the spectra of some integral operators and application to the numerical solution of some linear integral equations , author=. Applied Mathematics and Computation , volume=. 2011 , publisher=

    2011

  38. [38]

    SIAM review , volume=

    The spectral approximation of linear operators with applications to the computation of eigenelements of differential and integral operators , author=. SIAM review , volume=. 1981 , publisher=

    1981

  39. [39]

    2011 , publisher=

    Spectral approximation of linear operators , author=. 2011 , publisher=

    2011

  40. [40]

    SIAM Review , volume=

    Improvement of Rayleigh--Ritz eigenfunctions , author=. SIAM Review , volume=. 1972 , publisher=

    1972

  41. [41]

    Abstract and Applied Analysis , volume=

    Numerical methods for solving Fredholm integral equations of second kind , author=. Abstract and Applied Analysis , volume=. 2013 , organization=

    2013

  42. [42]

    The Journal of Analysis , volume=

    Galerkin methods for eigenvalue problem of an integral operator with logarithmic kernel , author=. The Journal of Analysis , volume=. 2017 , publisher=

    2017

  43. [43]

    Advances in Computational Mathematics , volume=

    A degenerate kernel method for eigenvalue problems of compact integral operators , author=. Advances in Computational Mathematics , volume=. 2007 , publisher=

    2007

  44. [44]

    2001 , publisher=

    Spectral computations for bounded operators , author=. 2001 , publisher=

    2001

  45. [45]

    International Conference on Machine Learning , pages=

    Beyond ucb: Optimal and efficient contextual bandits with regression oracles , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  46. [46]

    International Conference on Machine Learning , pages=

    Practical contextual bandits with regression oracles , author=. International Conference on Machine Learning , pages=. 2018 , organization=

    2018

  47. [47]

    The collected works of Wassily Hoeffding , pages=

    Probability inequalities for sums of bounded random variables , author=. The collected works of Wassily Hoeffding , pages=. 1994 , publisher=

    1994

  48. [48]

    Knowledge-Based Systems , volume=

    A novel risk-control model for the online portfolio selection of high-frequency transactions , author=. Knowledge-Based Systems , volume=. 2022 , publisher=

    2022

  49. [49]

    arXiv preprint arXiv:1310.0039 , year=

    Changing the paradigm of fixed significance levels: Testing Hypothesis by Minimizing Sum of Errors Type I and Type II , author=. arXiv preprint arXiv:1310.0039 , year=

    Pith/arXiv arXiv

  50. [50]

    THE ANNALS of STATISTICS , pages=

    ACTIVE SEQUENTIAL HYPOTHESIS TESTING , author=. THE ANNALS of STATISTICS , pages=. 2013 , publisher=

    2013

  51. [51]

    Journal of Mathematical Analysis and Applications , volume=

    Optimal decay rates in Sobolev norms for singular values of integral operators , author=. Journal of Mathematical Analysis and Applications , volume=. 2024 , publisher=

    2024

  52. [52]

    2023 , school=

    On the decay rate of singular values of integral operators , author=. 2023 , school=

    2023

  53. [53]

    Positivity , volume=

    Approximation tools and decay rates for eigenvalues of integral operators on a general setting , author=. Positivity , volume=. 2020 , publisher=

    2020

  54. [54]

    Advances in Neural Information Processing Systems , volume=

    Eigenvalue decay implies polynomial-time learnability for neural networks , author=. Advances in Neural Information Processing Systems , volume=

  55. [55]

    arXiv preprint arXiv:1805.08268 , year=

    Effective Dimension of Exp-concave Optimization , author=. arXiv preprint arXiv:1805.08268 , year=

    Pith/arXiv arXiv

  56. [56]

    2011 , publisher=

    Elements of operator theory , author=. 2011 , publisher=

    2011

  57. [57]

    2015 , publisher=

    Operator theory , author=. 2015 , publisher=

    2015

  58. [58]

    American Mathematical Society , year=

    A Course in Operator Theory , author=. American Mathematical Society , year=

  59. [59]

    1991 , publisher=

    A course in econometrics , author=. 1991 , publisher=

    1991

  60. [60]

    arXiv preprint arXiv:2205.14545 , year=

    Functional linear regression of cumulative distribution functions , author=. arXiv preprint arXiv:2205.14545 , year=

    Pith/arXiv arXiv

  61. [61]

    Translations of Mathematical Monographs , year=

    Convex Analysis: Theory and Applications , author=. Translations of Mathematical Monographs , year=

  62. [62]

    2014 , publisher=

    Understanding machine learning: From theory to algorithms , author=. 2014 , publisher=

    2014

  63. [63]

    2021 , publisher=

    Mathematical foundations of infinite-dimensional statistical models , author=. 2021 , publisher=

    2021

  64. [64]

    International Conference on Machine Learning , pages=

    Improved sleeping bandits with stochastic action sets and adversarial rewards , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  65. [65]

    Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics , pages=

    Contextual bandit algorithms with supervised learning guarantees , author=. Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics , pages=. 2011 , organization=

    2011

  66. [66]

    arXiv preprint arXiv:2211.06516 , year=

    Bandits for Online Calibration: An Application to Content Moderation on Social Media Platforms , author=. arXiv preprint arXiv:2211.06516 , year=

    Pith/arXiv arXiv

  67. [67]

    arXiv preprint arXiv:2101.01739 , year=

    Online multivalid learning: Means, moments, and prediction intervals , author=. arXiv preprint arXiv:2101.01739 , year=

    Pith/arXiv arXiv

  68. [68]

    Advances in Neural Information Processing Systems , volume=

    Practical adversarial multivalid conformal prediction , author=. Advances in Neural Information Processing Systems , volume=

  69. [69]

    2020 IEEE Congress on Evolutionary Computation (CEC) , pages=

    Survey on applications of multi-armed and contextual bandits , author=. 2020 IEEE Congress on Evolutionary Computation (CEC) , pages=. 2020 , organization=

    2020

  70. [70]

    arXiv preprint arXiv:1508.03326 , year=

    A survey on contextual multi-armed bandits , author=. arXiv preprint arXiv:1508.03326 , year=

    Pith/arXiv arXiv

  71. [71]

    Mobile health: sensors, analytic methods, and applications , pages=

    From ads to interventions: Contextual bandits in mobile health , author=. Mobile health: sensors, analytic methods, and applications , pages=. 2017 , publisher=

    2017

  72. [72]

    Information Systems Research , volume=

    Spoiled for choice? Personalized recommendation for healthcare decisions: A multiarmed bandit approach , author=. Information Systems Research , volume=. 2023 , publisher=

    2023

  73. [73]

    International conference on machine learning , pages=

    Taming the monster: A fast and simple algorithm for contextual bandits , author=. International conference on machine learning , pages=. 2014 , organization=

    2014

  74. [74]

    Annual Review of Statistics and Its Application , volume=

    Functional regression , author=. Annual Review of Statistics and Its Application , volume=. 2015 , publisher=

    2015

  75. [75]

    2002 , publisher=

    Applied functional data analysis: methods and case studies , author=. 2002 , publisher=

    2002

  76. [76]

    International Conference on Machine Learning , pages=

    Contextual bandits with large action spaces: Made practical , author=. International Conference on Machine Learning , pages=. 2022 , organization=

    2022

  77. [77]

    Conference on learning theory , pages=

    Provably efficient reinforcement learning with linear function approximation , author=. Conference on learning theory , pages=. 2020 , organization=

    2020

  78. [78]

    Sparse Contextual CDF Regression , author=

  79. [79]

    arXiv preprint arXiv:2007.07876 , year=

    Upper counterfactual confidence bounds: a new optimism principle for contextual bandits , author=. arXiv preprint arXiv:2007.07876 , year=

    Pith/arXiv arXiv 2007

  80. [80]

    arXiv preprint arXiv:2401.08016 , year=

    Contextual Bandits with Stage-wise Constraints , author=. arXiv preprint arXiv:2401.08016 , year=

    Pith/arXiv arXiv

Showing first 80 references.