Pointwise confidence estimation in the non-linear $\ell^2$-regularized least squares

Ilja Kuzborskij; Yasin Abbasi Yadkori

REVIEW 4 major objections 4 minor 43 references

Pointwise confidence estimation in the non-linear $\ell^2$-regularized least squares

T0 review · 4 major / 4 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 For ℓ2-regularized non-linear least squares, this paper proves a non-asymptotic pointwise confidence bound whose width is controlled by a Hessian-weighted gradient norm that grows outside the training data.

desk verdict A promising weighted-norm confidence idea for nonlinear least squares, but Lemma 4's sub-gamma argument is wrong and the main theorem does not follow as stated. read the letter →

arxiv 2506.07088 v2 pith:IWF7NERJ submitted 2025-06-08 cs.LG stat.ML

Ilja Kuzborskij , Yasin Abbasi Yadkori This is my paper

classification cs.LGstat.ML

keywords pointwiseconfidenceestimationnon-linearleastsquaresℓ2regularizationsub-gammaconcentrationweightednorminverseHessianfixeddesignregressionuncertaintyquantification

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

To know how much to trust a learned non-linear predictor at a single test point, one needs a confidence interval that does not rely on asymptotic arguments. This paper derives such an interval for $\ell^2$-regularized non-linear least squares under Gaussian noise: for any fixed test input $x$, the prediction error is controlled by a weighted norm of the predictor's gradient, with a weight matrix built from the regularized Hessian and the empirical Fisher information. The norm plays the role of $x^{\top} \Sigma^{-1} x$ from linear ridge regression, so the interval automatically widens when $x$ is poorly represented by the training data. The result holds for any local minimizer of the regularized loss and comes with an efficient conjugate-gradient algorithm that avoids forming the Hessian explicitly.

What carries the argument

The load-bearing object is the positive semidefinite matrix $M(\theta) = (\nabla^2 L(\theta) + \lambda I)^{-1} (1/n \sum_i \nabla_{\theta} f(x_i;\theta)\nabla_{\theta} f(x_i;\theta)^{\top}) (\nabla^2 L(\theta) + \lambda I)^{-1}$, a non-linear generalization of the inverse covariance that asymptotically approaches the inverse Fisher information. The proof differentiates the stationary-point condition to get $\partial \hat{\theta}/\partial z_i = (1/n)(\lambda I + \nabla^2 L(\hat{\theta}))^{-1} \nabla_{\theta} f(x_i;\hat{\theta})$, which turns the prediction's sensitivity to each noise variable into the weighted norm of the gradient. A Gaussian concentration inequality for differentiable functions of normal vectors then controls $|f(x;\hat{\theta}) - E[f(x;\hat{\theta})]|$ by the expected norm, and a sub-gamma tail assumption on the norm converts that into the explicit confidence interval. Computationally, the norm is evaluated by conjugate gradient with Hessian-vector products from automatic differentiation, at cost $\tilde{O}((n+p)/\sqrt{\lambda} + np)$ rather than $O(p^3)$.

What would settle it

Resample the noise many times for a fixed model and dataset, compute the centered weighted norm $W = \|\nabla_{\theta} f(x;\hat{\theta})\|^2_{M(\hat{\theta})} - E[\|\nabla_{\theta} f(x;\hat{\theta})\|^2_{M(\hat{\theta})}]$ for each draw, and check whether the empirical moment-generating function satisfies the sub-gamma inequality $\ln E[e^{\lambda W}] \le \lambda^2 v^2/(2(1-c\lambda))$ for the chosen $v,c$ and all $\lambda \in (0,1/c)$; if it fails, or if the empirical coverage of the interval over many datasets falls persistently below the nominal $1-\delta$, the load-bearing premise collapses.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1: if the centered weighted norm $\|\nabla_{\theta} f(x;\hat{\theta})\|^2_{M(\hat{\theta})}$ is sub-gamma with variance factor $v^2$ and scale $c$, then for every fixed test input $x$ and every $\delta \in (0,1)$, with probability at least $1-\delta$, $|f(x;\hat{\theta}) - E[f(x;\hat{\theta})]| \le \sigma \sqrt{(\pi^2/2) \ln(2/\delta) / n \cdot E[\|\nabla_{\theta} f(x;\hat{\theta})\|^2_{M(\hat{\theta})}]} + \sigma^2 \sqrt{2 \ln(2/\delta) v + (2/3) \ln(2/\delta) c}/n$, where $M(\theta) = (\nabla^2 L(\theta) + \lambda I)^{-1} (1/n \sum_i \nabla_{\theta} f(x_i;\theta) \nabla_{\theta} f(x_i;\theta)^{\top}) (\nabla^2 L(\theta) + \lambda I)^{-1}$. The bound requires only that the regularized loss be differentiable and that $\hat{\theta}$ be a local minimizer; it recovers the ridge-regression interval as a sanity check, its asymptotic shape matches the delta-method interval for the MLE, and the weighted norm is computable without ever forming the $p \times p$ inverse Hessian.

Load-bearing premise

The guarantee rests on the centered weighted norm $\|\nabla_{\theta} f(x;\hat{\theta})\|^2_{M(\hat{\theta})}$ being sub-gamma with known variance factor $v^2$ and scale $c$; the paper gives no way to estimate or certify $v$ and $c$, and the experiments simply set both to 1, so the announced $\delta$-level coverage is conditional on an unverified tail assumption.

Editorial extensions

If this is right

A single trained non-linear model yields pointwise confidence bands that automatically widen in regions with few or no training observations; in the paper's experiments this is the behavior that bootstrap-based intervals fail to deliver.
For linear models the bound reduces to the standard ridge-regression interval with the known $x^{\top}\Sigma^{-1} x$ weighting, so the non-linear result is a strict generalization of a classical guarantee.
Under consistency the bound's asymptotic form matches the delta-method interval from asymptotic normality of the MLE, giving a finite-sample analogue rather than a separate heuristic.
Uniform confidence over a whole input sphere follows for ReLU networks with only a mild covering-factor cost once the training procedure controls the predictor's Lipschitz constant.
The weighted norm is computed by conjugate gradient with complexity roughly $\tilde{O}((n+p)/\sqrt{\lambda} + np)$ and $O(p)$ memory, so it scales to large parameter counts.

Reading between the lines

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

If $v$ and $c$ were estimated from data (for example by subsampling or empirical tail bounds) and plugged into the interval, the $\delta$-level guarantee would become fully empirical; the paper leaves that as an open engineering step.
The polynomial-tail variant in the appendix removes the sub-gamma assumption at the cost of $1/\delta$ rather than $\ln(1/\delta)$, so a practical implementation could switch between the two bounds depending on whether a certified tail is available.
Because ReLU networks are not differentiable everywhere, the formal theorem does not literally cover the experiments; a smoothed-activation or subgradient version of the proof would likely preserve the same interval shape and is a natural next step.
Testable extension: for a dataset with known ground truth, estimate the empirical coverage of the interval with $v=c=1$ across many noise draws; the gap from the nominal level quantifies how sensitive the guarantee is to the unverified sub-gamma parameters.

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.

Desk Editor's Note

read the letter

Quick take: the paper has a genuinely appealing idea — a pointwise non-asymptotic confidence bound for local minimizers of ℓ2-regularized nonlinear least squares, with an inverse-Hessian weighted norm that grows outside the training data. The efficient CG/HVP computation is a real practical contribution, and the discussion connecting the bound to Fisher information and the NTK is informative. The reader's conditional verdict is more charitable than the text supports.

The soft spots are serious. The proof of the main theorem goes through Lemma 4, and that lemma's proof applies the sub-gamma mgf at the wrong argument. Pisier's inequality gives E exp(λ|F−EF|) ≤ 2 E exp((λ²π²/8)||∇F||²). If X=||∇F||² is sub-gamma with variance factor v² and scale c, its mgf at argument t=λ²π²/8 yields an exponent of order λ⁴v²/(1−cλ²), not the λ²(v²/2)/(1−cλ) the paper uses. The theorem's second term, with its √(2 ln(2/δ)v + …), is therefore not established even if v and c were known exactly. This is a load-bearing proof gap, not a tuning issue.

The experiments do not repair it. They set v=c=1 without estimation, which is a heuristic, and the claimed δ-level guarantee is not delivered in practice. They also evaluate coverage against the true function, while the theorem controls only the variance around E[f(x;θhat)]. And the ReLU networks used in the experiments are outside the assumptions: Lλ is not twice differentiable with invertible regularized Hessian at the solution.

What is actually new: the weighted-norm representation of uncertainty, the non-asymptotic formulation, and an efficient way to compute the norm. These are worth taking seriously. The paper reads like an honest attempt, and the error is localized and likely fixable — a corrected Lemma 4 may give a weaker bound or require a different concentration argument.

Who this is for: researchers in UQ for neural networks who are looking for alternatives to asymptotic MLE/Laplace intervals. If the proof is repaired, it will be a useful paper. In its current form I would not cite it, but I would send it to a serious referee: the idea is important enough to justify chasing down the proof and forcing a clean revision.

Referee Report

4 major / 4 minor

Summary. The paper proposes a non-asymptotic, high-probability pointwise confidence bound for predictions of a local minimizer of an l2-regularized non-linear least-squares objective with fixed design and Gaussian noise. The bound is expressed through a weighted norm ||∇θ f(x; θhat)||²_{M(θhat)} involving the inverse regularized Hessian, which the authors connect to the Fisher information matrix and to classical asymptotic MLE confidence intervals. They also give a conjugate-gradient based algorithm for computing the weighted norm with cost close to that of a gradient evaluation, a uniform bound over the unit sphere, and experiments comparing the method with bootstrap confidence intervals. The central theoretical result is Theorem 1, whose proof relies on an implicit differentiation lemma and a concentration lemma (Lemma 4) based on Pisier's Gaussian concentration inequality and a sub-gamma tail assumption on the weighted norm.

Significance. If the main theorem were correct, the paper would make a useful contribution: it gives a concrete, computable uncertainty estimate that grows away from the training data, without requiring re-training as in bootstrap; the efficient Hessian-inverse-vector product computation is a practical strength; and the fixed-design, non-asymptotic formulation is a worthwhile complement to asymptotic MLE intervals. The paper also contains a clearly stated limitation of not controlling bias. However, these strengths are tied to the validity of Theorem 1, and the proof of that theorem has a load-bearing algebraic error that is not merely a presentation issue.

major comments (4)

[Section 4.2, Lemma 4] Lemma 4 is the proof engine for Theorem 1, and its derivation is incorrect. Pisier's inequality (Lemma 2) requires bounding E exp(λ²π²/8 ∥∇F(S)∥²), so the sub-gamma mgf from Definition 1 must be applied at argument t = λ²π²/8. That substitution gives an exponent of order λ⁴v²/(1 − cλ²π²/8), not λ²(v²/2)/(1 − cλ) as written in the proof. The subsequent Chernoff inversion therefore does not produce the term sqrt(2 ln(2/δ)v + (2/3)ln(2/δ)c)/n in Theorem 2; the stated finite-sample level is unsupported even when v and c are known exactly. The v/v² mismatch between Definition 1 and the statement of Theorem 2 is a symptom of the same error.
[Section 3, Theorem 1 and Section 5] The theorem conditions on unknown sub-gamma parameters v² and c, and the paper offers no estimation or certification procedure for them. Section 5 states that the proposed method is 'the bound in Theorem 1 with ν=c=1'; this does not yield the claimed δ-level guarantee unless the true parameters are known to be no larger than 1. Moreover, the empirical coverage is measured against the ground-truth function f(x;θ⋆), whereas Theorem 1 bounds |f(x;θhat)−E[f(x;θhat)]| and explicitly does not control the bias term E[f(x;θhat)]−f(x;θ⋆). The reported coverage numbers therefore cannot be read as a validation of the theorem's probability statement.
[Section 5 and Lemma 1] The experiments use ReLU networks trained with AdamW. Lemma 1, on which Theorem 1 rests, requires Lλ to be differentiable in θ and the regularized Hessian at the local minimizer to be invertible; ReLU networks are not differentiable at the activation kinks, and AdamW is not shown to reach an exact local minimizer. As a result, the experimental setting lies outside the theorem's assumptions, so the empirical claims about neural-network confidence intervals are not supported by the theory.
[Appendix C.3] The covering argument for the uniform bound over S^{d−1} treats bound(ln(1/δ)) as though it depends only on δ, but the bound in Theorem 1 depends on x through E∥∇θf(x;θhat)∥²_{M(θhat)}. Passing from the finite cover to the whole sphere requires a sup over x of that expectation, or a separate union bound for each x; this step is not written in the proof.

minor comments (4)

[Theorem 1 and Theorem 2] The σ-dependence of the second term is inconsistent: Theorem 1 has an extra σ² in front of the sub-gamma term while Theorem 2 does not, and M(θ) is defined without σ² even though the noise scale σ enters the gradient with respect to the noise variable. These definitions should be aligned.
[Definition 1 and Section 3] Definition 1 uses a variance factor v², but several theorem statements and the experimental section write v in the bound; the notation should be made consistent throughout.
[Section 3.1] The claimed CG iteration count k = Ω(sqrt(1 + B²/λ) ln(1/ε)) is stated without proof; please either prove it or describe it as a heuristic complexity estimate.
[Lemma 8 and Section 3.1] The Lipschitz bound in Lemma 8 contains (Lλ(θ)/(λK))^{K/2} and therefore requires λ > 0, so the discussion of the interpolation regime with λ = 0 in Section 3.1 should be explicitly qualified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the confidence bound is derived from a gradient-norm concentration argument, and the weighted norm is not constructed from the target gap it controls.

full rationale

The derivation is self-contained. Theorem 1 controls the point variance gap Δ(x;θhat)=f(x;θhat)−E[f(x;θhat)] using the weighted norm ||∇θf(x;θhat)||²_{M(θhat)}, which is a derivative-of-prediction statistic computed from the local minimizer and the empirical loss Hessian; it is not defined in terms of Δ or of the coverage event being bounded. Lemma 1 derives the identity ∂θhat/∂z_i = (1/n)(λI+∇²L(θhat))^{-1}∇θf(x_i;θhat), so ||∇_z f(x;θhat)||² equals the weighted norm, and Lemma 4 (via Pisier's inequality and the assumed sub-gamma tail on that norm) converts this into a concentration inequality for F(S)−E[F(S)]. No parameter is fitted to the target quantity: v and c are assumed tail constants, and the experiment's choice v=c=1 is a calibration step, not a fit to the predicted coverage. The only self-citation (Yadkori, 2012) appears in the related-work discussion of kernel confidence intervals and is not load-bearing; no uniqueness theorem or ansatz is imported from the authors' prior work. The linear-regression sanity check recovers a known ridge bound but is not used as the definition of the nonlinear result. Any concern about the algebra in Lemma 4 is a proof-correctness issue, not circularity, because even a gap there would not make the conclusion equivalent to the assumption by construction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The theorem's guarantee is conditional on Gaussian noise with known σ, differentiability of Lλ at θhat, and a sub-gamma assumption on the data-dependent weighted norm. The paper does not provide a data-dependent estimator for v and c, and the experiments fix them to 1. The reported intervals also implicitly assume small bias when compared to the true function.

free parameters (3)

sub-gamma variance factor v^2 = 1 (set in experiments)
Theorem 1 requires this parameter; no estimation procedure is given, and the experiments set v=1.
sub-gamma scale c = 1 (set in experiments)
Same as v; c controls the linear tail term in the bound and is not estimated.
regularization strength λ = 10^-8 to 10^-3 depending on input dimension, tuned on validation data
Regularization enters the Hessian and the matrix M(θ); it is a hyperparameter chosen by the practitioner rather than fitted to the target result.

assumptions (4)

domain assumption Responses follow Yi = f(xi;θ*) + σZi with Zi standard Gaussian and σ known.
Used in Eq. (2) and in Pisier's Gaussian concentration inequality, which is central to the proof.
domain assumption Lλ is differentiable and θhat is a local minimizer; θhat(z) is differentiable in each noise coordinate.
Needed for Lemma 1, the implicit differentiation step that connects the prediction gap to the weighted norm. ReLU networks violate this at activation boundaries.
ad hoc to paper The centered weighted norm is sub-gamma with known variance factor v^2 and scale c.
Assumed in Theorem 1; no estimator is provided, and experiments set v=c=1.
domain assumption For confidence intervals targeting f(x;θ*), the bias E[f(x;θhat)]-f(x;θ*) is negligible.
The paper explicitly says bias is not controlled, yet the experiments evaluate coverage of the ground truth f(x;θ*).

how reviews work

0 comments

Cite this review

Pith. "Pith review of Pointwise confidence estimation in the non-linear $\ell^2$-regularized least squares." pith.science (2026). https://pith.science/paper/IWF7NERJ

@misc{pith2026250607088,
  author       = {Pith},
  title        = {Pith review of: Pointwise confidence estimation in the non-linear $\ell^2$-regularized least squares},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IWF7NERJ}},
  note         = {Machine review of arXiv:2506.07088}
}

abstract

We consider a high-probability non-asymptotic confidence estimation in the $\ell^2$-regularized non-linear least-squares setting with fixed design. In particular, we study confidence estimation for local minimizers of the regularized training loss. We show a pointwise confidence bound, meaning that it holds for the prediction on any given fixed test input $x$. Importantly, the proposed confidence bound scales with similarity of the test input to the training data in the implicit feature space of the predictor (for instance, becoming very large when the test input lies far outside of the training data). This desirable last feature is captured by the weighted norm involving the inverse-Hessian matrix of the objective function, which is a generalized version of its counterpart in the linear setting, $x^{\top} \text{Cov}^{-1} x$. Our generalized result can be regarded as a non-asymptotic counterpart of the classical confidence interval based on asymptotic normality of the MLE estimator. We propose an efficient method for computing the weighted norm, which only mildly exceeds the cost of a gradient computation of the loss function. Finally, we complement our analysis with empirical evidence showing that the proposed confidence bound provides better coverage/width trade-off compared to a confidence estimation by bootstrapping, which is a gold-standard method in many applications involving non-linear predictors such as neural networks.

Figures

Figures reproduced from arXiv: 2506.07088 by the authors.

**Figure 2.** Trade-off between CI width and coverage, when varying regularization parameter [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

43 extracted references · 33 canonical work pages

[1]

Adapting the linearised laplace model evidence for modern deep learning

Javier Antor \'a n, David Janz, James U Allingham, Erik Daxberger, Riccardo Rb Barbano, Eric Nalisnick, and Jos \'e Miguel Hern \'a ndez-Lobato. Adapting the linearised laplace model evidence for modern deep learning. In International Conference on Machine Learing (ICML), 2022

work page 2022
[2]

Model-based reinforcement learning with value-targeted regression

Alex Ayoub, Zeyu Jia, Csaba Szepesvari, Mengdi Wang, and Lin Yang. Model-based reinforcement learning with value-targeted regression. In International Conference on Machine Learing (ICML), 2020

work page 2020
[3]

Weight uncertainty in neural network

Charles Blundell, Julien Cornebise, Koray Kavukcuoglu, and Daan Wierstra. Weight uncertainty in neural network. In International Conference on Machine Learing (ICML), 2015

work page 2015
[4]

Concentration inequalities using the entropy method

St \'e phane Boucheron, G \'a bor Lugosi, and Pascal Massart. Concentration inequalities using the entropy method. The Annals of Probability, 31 0 (3): 0 1583--1614, 2003

2003
[5]

Concentration inequalities: A nonasymptotic theory of independence

St \'e phane Boucheron, G \'a bor Lugosi, and Pascal Massart. Concentration inequalities: A nonasymptotic theory of independence. Oxford University Press, 2013

work page 2013
[6]

On kernelized multi-armed bandits

Sayak Ray Chowdhury and Aditya Gopalan. On kernelized multi-armed bandits. In International Conference on Machine Learing (ICML), pages 844--853. PMLR, 2017

work page 2017
[7]

Distribution-free uncertainty quantification for kernel methods by gradient perturbations

Bal \'a zs Csan \'a d Cs \'a ji and B \'a lint Horv \'a th. Distribution-free uncertainty quantification for kernel methods by gradient perturbations. Machine Learning, 108 0 (8): 0 1677--1699, 2019

work page 2019
[8]

A simple convergence proof of adam and adagrad

Alexandre D \'e fossez, L \'e on Bottou, Francis R Bach, and Nicolas Usunier. A simple convergence proof of adam and adagrad. Transactions on Machine Learning Research (TMLR), 2022

work page 2022

Show all 43 references

[9]

Ensembles for uncertainty estimation: Benefits of prior functions and bootstrapping

Vikranth Dwaracherla, Zheng Wen, Ian Osband, Xiuyuan Lu, Seyed Mohammad Asghari, and Benjamin Van Roy. Ensembles for uncertainty estimation: Benefits of prior functions and bootstrapping. arXiv preprint arXiv:2206.03633, 2022

2022 arXiv
[10]

An introduction to the bootstrap

Bradley Efron and Robert J Tibshirani. An introduction to the bootstrap. Chapman and Hall/CRC, 1994

1994
[11]

'in-between' uncertainty in bayesian neural networks

Andrew YK Foong, Yingzhen Li, Jos \'e Miguel Hern \'a ndez-Lobato, and Richard E Turner. 'in-between' uncertainty in bayesian neural networks. arXiv preprint arXiv:1906.11537, 2019

1906 arXiv
[12]

Practical contextual bandits with regression oracles

Dylan Foster, Alekh Agarwal, Miroslav Dudik, Haipeng Luo, and Robert Schapire. Practical contextual bandits with regression oracles. In International Conference on Machine Learing (ICML), pages 1539--1548. PMLR, 2018

2018
[13]

Dropout as a bayesian approximation: Representing model uncertainty in deep learning

Yarin Gal and Zoubin Ghahramani. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In International Conference on Machine Learing (ICML), 2016

2016
[14]

Stochastic first- and zeroth-order methods for nonconvex stochastic programming

Saeed Ghadimi and Guanghui Lan. Stochastic first- and zeroth-order methods for nonconvex stochastic programming. SIAM Journal on Optimization, 23 0 (4): 0 2341--2368, 2013

2013
[15]

A distribution-free theory of nonparametric regression

L \'a szl \'o Gy \"o rfi, Michael Kohler, Adam Krzyzak, and Harro Walk. A distribution-free theory of nonparametric regression. Springer Science & Business Media, 2006

2006
[16]

Inequalities on the lambert w function and hyperpower function

Abdolhossein Hoorfar and Mehdi Hassani. Inequalities on the lambert w function and hyperpower function. J. Inequal. Pure and Appl. Math, 9 0 (2): 0 5--9, 2008

2008
[17]

Improving predictions of bayesian neural nets via local linearization

Alexander Immer, Maciej Korzepa, and Matthias Bauer. Improving predictions of bayesian neural nets via local linearization. In Arindam Banerjee and Kenji Fukumizu, editors, International Conference on Artificial Intelligence and Statistics (AISTATS), 2021

2021
[18]

Neural tangent kernel: Convergence and generalization in neural networks

Arthur Jacot, Franck Gabriel, and Cl \'e ment Hongler. Neural tangent kernel: Convergence and generalization in neural networks. Advances in Neural Information Processing Systems, 31, 2018

2018
[19]

Online sub-sampling for reinforcement learning with general function approximation

Dingwen Kong, Ruslan Salakhutdinov, Ruosong Wang, and Lin F Yang. Online sub-sampling for reinforcement learning with general function approximation. In Workshop on Reinforcement Learning Theory at ICML, 2021

2021
[20]

Simple and scalable predictive uncertainty estimation using deep ensembles

Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. Simple and scalable predictive uncertainty estimation using deep ensembles. Advances in Neural Information Processing Systems, 2017

2017
[21]

A lower bound for linear and kernel regression with adaptive covariates

Tor Lattimore. A lower bound for linear and kernel regression with adaptive covariates. In Conference on Computational Learning Theory (COLT), 2023

2023
[22]

Bandit algorithms

Tor Lattimore and Csaba Szepesv \'a ri. Bandit algorithms. Cambridge University Press, 2020

2020
[23]

Convergence of adam under relaxed assumptions

Haochuan Li, Alexander Rakhlin, and Ali Jadbabaie. Convergence of adam under relaxed assumptions. Advances in Neural Information Processing Systems, 2023

2023
[24]

Decoupled weight decay regularization

Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In International Conference on Learning Representations (ICLR), 2019

2019
[25]

Bayesian Methods for Adaptive Models

David John Cameron Mackay. Bayesian Methods for Adaptive Models. PhD thesis, California Institute of Technology, USA, 1992

1992
[26]

Stochastic gradient markov chain monte carlo

Christopher Nemeth and Paul Fearnhead. Stochastic gradient markov chain monte carlo. Journal of the American Statistical Association, 116 0 (533): 0 433--450, 2021

2021
[27]

Deep exploration via bootstrapped dqn

Ian Osband, Charles Blundell, Alexander Pritzel, and Benjamin Van Roy. Deep exploration via bootstrapped dqn. Advances in Neural Information Processing Systems, 2016

2016
[28]

Randomized prior functions for deep reinforcement learning

Ian Osband, John Aslanides, and Albin Cassirer. Randomized prior functions for deep reinforcement learning. Advances in Neural Information Processing Systems, 2018

2018
[29]

Epistemic neural networks

Ian Osband, Zheng Wen, Seyed Mohammad Asghari, Vikranth Dwaracherla, Morteza Ibrahimi, Xiuyuan Lu, and Benjamin Van Roy. Epistemic neural networks. Advances in Neural Information Processing Systems, 2023

2023
[30]

Fast exact multiplication by the hessian

Barak A Pearlmutter. Fast exact multiplication by the hessian. Neural computation, 6 0 (1): 0 147--160, 1994

1994
[31]

Probabilistic methods in the geometry of banach spaces

Gilles Pisier. Probabilistic methods in the geometry of banach spaces. Lecture Notes in Mathematics, pages 167--241, 1986

1986
[32]

Carl Edward Rasmussen and Christopher K. I. Williams. Gaussian Processes for Machine Learning. MIT press, 2006

2006
[33]

Learning to optimize via posterior sampling

Daniel Russo and Benjamin Van Roy. Learning to optimize via posterior sampling. Mathematics of Operations Research, 39 0 (4): 0 1221--1243, 2014

2014
[34]

Understanding machine learning: From theory to algorithms

Shai Shalev-Shwartz and Shai Ben-David. Understanding machine learning: From theory to algorithms. Cambridge university press, 2014

2014
[35]

Mathematical statistics

Jun Shao. Mathematical statistics. Springer, 2003

2003
[36]

Gaussian process optimization in the bandit setting: no regret and experimental design

Niranjan Srinivas, Andreas Krause, Sham Kakade, and Matthias Seeger. Gaussian process optimization in the bandit setting: no regret and experimental design. In International Conference on Machine Learing (ICML), 2010

2010
[37]

An introduction to matrix concentration inequalities

Joel A Tropp et al. An introduction to matrix concentration inequalities. Foundations and Trends in Machine Learning , 8 0 (1-2): 0 1--230, 2015

2015
[38]

Asymptotic statistics, volume 3

Aad W Van der Vaart. Asymptotic statistics, volume 3. Cambridge University Press, 2000

2000
[39]

High-dimensional statistics: A non-asymptotic viewpoint, volume 48

Martin J Wainwright. High-dimensional statistics: A non-asymptotic viewpoint, volume 48. Cambridge University Press, 2019

2019
[40]

All of statistics: a concise course in statistical inference, volume 26

Larry A Wasserman. All of statistics: a concise course in statistical inference, volume 26. Springer, 2004

2004
[41]

A decision-theoretic approach to interval estimation

Robert L Winkler. A decision-theoretic approach to interval estimation. Journal of the American Statistical Association, 67 0 (337): 0 187--191, 1972

1972
[42]

Online Learning for Linearly Parametrized Control Problems

Yasin Abbasi Yadkori. Online Learning for Linearly Parametrized Control Problems. Phd, University Of Alberta, 2012

2012
[43]

Towards understanding convergence and generalization of adamw

Pan Zhou, Xingyu Xie, Zhouchen Lin, and Shuicheng Yan. Towards understanding convergence and generalization of adamw. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

2024

Pith tools

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

[1] [1]

Adapting the linearised laplace model evidence for modern deep learning

Javier Antor \'a n, David Janz, James U Allingham, Erik Daxberger, Riccardo Rb Barbano, Eric Nalisnick, and Jos \'e Miguel Hern \'a ndez-Lobato. Adapting the linearised laplace model evidence for modern deep learning. In International Conference on Machine Learing (ICML), 2022

work page 2022

[2] [2]

Model-based reinforcement learning with value-targeted regression

Alex Ayoub, Zeyu Jia, Csaba Szepesvari, Mengdi Wang, and Lin Yang. Model-based reinforcement learning with value-targeted regression. In International Conference on Machine Learing (ICML), 2020

work page 2020

[3] [3]

Weight uncertainty in neural network

Charles Blundell, Julien Cornebise, Koray Kavukcuoglu, and Daan Wierstra. Weight uncertainty in neural network. In International Conference on Machine Learing (ICML), 2015

work page 2015

[4] [4]

Concentration inequalities using the entropy method

St \'e phane Boucheron, G \'a bor Lugosi, and Pascal Massart. Concentration inequalities using the entropy method. The Annals of Probability, 31 0 (3): 0 1583--1614, 2003

2003

[5] [5]

Concentration inequalities: A nonasymptotic theory of independence

St \'e phane Boucheron, G \'a bor Lugosi, and Pascal Massart. Concentration inequalities: A nonasymptotic theory of independence. Oxford University Press, 2013

work page 2013

[6] [6]

On kernelized multi-armed bandits

Sayak Ray Chowdhury and Aditya Gopalan. On kernelized multi-armed bandits. In International Conference on Machine Learing (ICML), pages 844--853. PMLR, 2017

work page 2017

[7] [7]

Distribution-free uncertainty quantification for kernel methods by gradient perturbations

Bal \'a zs Csan \'a d Cs \'a ji and B \'a lint Horv \'a th. Distribution-free uncertainty quantification for kernel methods by gradient perturbations. Machine Learning, 108 0 (8): 0 1677--1699, 2019

work page 2019

[8] [8]

A simple convergence proof of adam and adagrad

Alexandre D \'e fossez, L \'e on Bottou, Francis R Bach, and Nicolas Usunier. A simple convergence proof of adam and adagrad. Transactions on Machine Learning Research (TMLR), 2022

work page 2022