Outlier-Robust Linear System Identification Under Heavy-tailed Noise

Aritra Mitra; Vinay Kanakeri

REVIEW 2 major objections 3 minor 1 cited by

Outlier-Robust Linear System Identification Under Heavy-tailed Noise

T0 review · 2 major / 3 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Even when a linear system's process noise has only a finite fourth moment and arbitrarily heavy tails, a bucketed least-squares estimate fused by a geometric median recovers nearly the same sample-complexity bound as Gaussian noise, with…

desk verdict Theorems 1–2 are a genuine and likely correct advance for heavy-tailed sys-ID; Theorem 3's adversarial guarantee is not proven as written, so the paper needs a revision before the robustness claim can stand. read the letter →

arxiv 2501.00421 v2 pith:NAR5APQV submitted 2024-12-31 eess.SY cs.LGcs.SYmath.OC

Vinay Kanakeri , Aritra Mitra This is my paper · ORCID

classification eess.SYcs.LGcs.SYmath.OC MSC 93E1262F3562H12

keywords heavy-tailednoiselinearsystemidentificationfinitefourthmomentgeometricmedianofmeanssamplecomplexityPACboundsadversarialcorruption

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 asks whether the transition matrix of a linear time-invariant system can be learned when the process noise is heavy-tailed, with only a finite fourth moment and no sub-Gaussian tails. The authors propose a robust estimator, Robust-SysID, that splits the collected trajectories into buckets, computes an ordinary least-squares estimate in each bucket, and then fuses the bucket estimates with a geometric median. They prove that this estimator recovers the same logarithmic dependence on the failure probability as the sub-Gaussian OLS benchmark: exactly in the scalar case, and up to an extra factor of $d^{3/2}$ in the vector case. The number of trajectories must grow with the kurtosis of the noise distribution, which quantifies the price of heavy tails. The same algorithm is shown to tolerate adversarial corruption of a small fraction of trajectories.

What carries the argument

The load-bearing mechanism is the combination of bucketing with the geometric median: within each bucket a plain least-squares fit is only weakly concentrated (failure probability decays polynomially), but the median across $K=\lceil c_1\log(1/\delta)\rceil$ independent buckets boosts the failure probability to $\delta$ via Hoeffding's inequality. The vector analysis rests on a fourth-moment identity for $\mathbb{E}[(x_T x_T^\top)^2]$ (Lemma 5) that lets one bound the trace of the whitened residual covariance, yielding the lower bound $\lambda_{\min}(\sum_{i\in B_j} z_T^{(i)}(z_T^{(i)})^\top)\ge M/2$ once each bucket holds at least $M\ge c d^2 C_A C_w$ trajectories. This denominator bound, together with a Frobenius-norm Markov bound on the numerator $\sum_{i\in B_j} w_T^{(i)}(z_T^{(i)})^\top$, is what replaces the sub-Gaussian concentration arguments of prior work.

What would settle it

Simulate the scalar system with a zero-mean noise distribution that has a finite fourth moment but a deliberately large kurtosis (for example, a $t$-distribution with five degrees of freedom, scaled to variance $\sigma^2$), set $K=\lceil c_1\log(1/\delta)\rceil$ and $M=c_2\tilde\sigma^4/\sigma^4$, and check empirically whether the failure probability of $|\hat a-a|>C\sqrt{\log(1/\delta)/(N g_T)}$ stays below $\delta$; a systematic violation would refute the claimed bound.

Watch

Extended reading notes

Core claim

The central claim is that a median-of-means style construction eliminates the need for light-tailed noise in finite-sample linear system identification. Under the noise model $\mathbb{E}[w_t w_t^\top]=\sigma^2 I$ and $\mathbb{E}[(w_t(i))^4]=\tilde\sigma^4$, Robust-SysID achieves, with probability at least $1-\delta$, the bound $\|\hat{A}-A\| \le C d^{3/2}\sqrt{\log(1/\delta)/(N\lambda_{\min}(G_T))}$ when each bucket contains $M\ge c_2 d^2 C_A C_w$ trajectories and the number of buckets is $K=\lceil c_1\log(1/\delta)\rceil$. In the scalar case the bound becomes $|\hat a-a|\le C\sqrt{\log(1/\delta)/(N g_T)}$, identical in form to the sub-Gaussian OLS rate. The proof avoids all sub-Gaussian concentration arguments, relying instead on Chebyshev and Markov inequalities together with a fourth-moment computation that controls the smallest eigenvalue of the whitened empirical covariance. A corollary is that when an adversary corrupts an $\eta$ fraction of trajectories, the same estimator's error degrades only by an additive $O(\sqrt{\eta})$ term.

Load-bearing premise

The entire proof requires the noise to be zero-mean with isotropic covariance ($\mathbb{E}[w_t w_t^\top]=\sigma^2 I$), a finite fourth moment per coordinate, and every trajectory to start from the zero state, so that the state covariance is exactly $\sigma^2 G_T$ and the whitening step is valid.

Editorial extensions

If this is right

Under noise with only a finite fourth moment, the failure probability enters the estimator's error bound logarithmically, so high-confidence guarantees are not much costlier than under Gaussian noise.
In the scalar case the robust estimator matches the sub-Gaussian OLS rate exactly; in the vector case it matches up to a factor $d^{3/2}$ in the error bound and $O(d^2)$ in the required number of trajectories.
The number of trajectories needed per bucket grows linearly with the kurtosis $\tilde\sigma^4/\sigma^4$, making the cost of heavy tails explicit.
With adversarial corruption of up to an $\eta$ fraction of trajectories, the same estimator still delivers the heavy-tailed bound plus an additive $O(\sqrt{\eta/\lambda_{\min}(G_T)})$ error.
Under a controllability assumption the technique extends to systems with inputs, where the input matrix is also unknown.

Reading between the lines

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

The bucketed geometric-median construction is a generic booster: any least-squares type estimator whose error is bounded in probability with polynomial failure rate can be promoted to logarithmic failure probability, so the same recipe should apply to other identification and control tasks such as estimating the $B$ matrix or the Riccati solution.
The extra $d^{3/2}$ factor appears tied to the use of the Frobenius norm in the geometric-median step and to Markov-type concentration; if a spectral-norm geometric median with comparable guarantees existed, the vector bound might close to the sub-Gaussian rate, a gap the authors leave open.
A natural testable extension is whether a trimmed-mean or median-of-means variant that truncates by coordinate could reduce the kurtosis dependence below linear, since the paper only shows the linear dependence appears in its analysis, not that it is fundamental.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: Theorems 1 and 2 are the real content, and they look right. Theorem 3, the adversarial robustness extension, has a gap that needs fixing before the abstract's "arbitrarily corrupt" claim can be taken at face value.

What's genuinely new: the paper gives the first finite-sample PAC bounds for multi-trajectory LTI system identification under noise with only a finite fourth moment. The scalar case matches the sub-Gaussian rate exactly, and the vector case recovers it up to an O(d) factor. That is a real improvement over sub-Gaussian or sub-Weibull assumptions in prior work. The analysis is refreshingly elementary: Chebyshev on each bucket, a fourth-moment control of the covariance denominator, and a geometric-median boost. I checked the central fourth-moment trace bounds in Lemma 15 and they hold up; the whitening step is legitimate under the stated assumptions. The kurtosis-dependent trajectory requirement is an interesting and honest limitation, not hidden.

Where it gets soft: the stress-test note is correct. In Appendix D, the proof of Theorem 3 applies Hoeffding to Z_j W_j, where W_j indicates whether bucket j is uncorrupted. Under the strong-contamination model, the adversary can choose corruptions after seeing the data, so W_j is not independent of the bucket contents. The marginal bound E[Z_j W_j] ≤ p is derived by conditioning on W_j=1 and invoking the clean-bucket guarantee, but if the adversary corrupts buckets that look bad, that conditional probability can exceed p. The centered sum need not concentrate at the claimed rate, and the displayed exponent is not justified. This is a load-bearing gap in Theorem 3, not a typo. It does not damage Theorems 1–2.

One smaller caveat: the model assumes isotropic noise covariance and zero initial condition. Those are stated explicitly in Section 2, so I would not call them hidden assumptions, but the abstract oversells the generality by not mentioning the isotropic second-moment requirement. That is a presentational fix, not a mathematical one.

Verdict: send to peer review. The heavy-tailed results deserve referee time and the adversarial claim can be repaired or appropriately weakened. A serious referee should focus on Theorem 3 and the independence issue there. I would cite this paper for Theorems 1–2.

Referee Report

2 major / 3 minor

Summary. The paper studies finite-sample identification of the transition matrix A in the LTI system x_{t+1} = A x_t + w_t from N independent length-T trajectories, under a noise model (Eq. (2)) that assumes only zero mean, covariance σ²I, and finite fourth moments E[w_t(i)^4] = σ̃⁴. The proposed algorithm Robust-SysID partitions the trajectories into K buckets, forms an OLS estimate from each bucket using the last two states, and aggregates the bucket estimates with a Frobenius-norm geometric median. Theorem 1 gives a scalar error bound matching the sub-Gaussian OLS rate up to constants, with a kurtosis-dependent number of trajectories per bucket. Theorem 2 gives a vector bound ||Â − A|| ≤ C d^{3/2} sqrt(log(1/δ)/(N λ_min(G_T))) with K = ceil(c1 log(1/δ)) and M ≥ c2 d² C_A C_w. Theorem 3 claims an additive O(√η) error under an adversary that can corrupt an η fraction of trajectories. The proofs combine Chebyshev/Markov bounds on the per-bucket OLS numerator and denominator, followed by Hoeffding-based boosting through the median.

Significance. If Theorems 1 and 2 hold, the paper makes a solid contribution: it provides the first PAC-style system identification guarantees under a noise model that requires only a finite fourth moment, and it nearly matches the sub-Gaussian benchmark of Dean et al. [7], with only an extra O(d) factor in the vector setting. The scalar theorem, in particular, is clean and fully derived, and the explicit dependence on the kurtosis σ̃⁴/σ⁴ is a meaningful and informative feature. The proof strategy—combining weak per-bucket OLS estimators with geometric-median boosting—is natural and well executed for the clean-data setting. However, the adversarial extension in Theorem 3 is not proven as written: the Hoeffding argument in Appendix D requires an independence that the stated strong-contamination model does not provide. This gap weakens the advertised adversarial robustness claim, although it does not undermine the heavy-tailed identification results that form the paper's core.

major comments (2)

[Appendix D, Eq. (31)] The Hoeffding step in the proof of Theorem 3 is not justified under the stated strong-contamination model. The proof introduces W_j as the indicator that bucket j has no corruptions, and then centers Z_j W_j by E[Z_j W_j] and applies Hoeffding. But if the adversary may choose the corrupted trajectories after seeing the realized data—the standard strong-contamination model—then W_j is a random variable depending on the clean data and on the other W_j's. Consequently the summands Z_j W_j − E[Z_j W_j] are not independent, and Hoeffding's inequality cannot be applied as written. Moreover, the bound E[Z_j W_j] ≤ p derived at the end of Appendix D uses (29), which is an unconditional guarantee for a clean bucket; conditioning on W_j = 1 can select unfavorable realizations of the clean data, so P({||Â_j − A|| ≥ ε} ∩ {W_j = 1}) need not be ≤ p. Thus the displayed probability bound leading to the statement of Theorem 3 is not established. This is a load-bearing gap in the adversarial contribution, although it does not invalidate Theorems 1 and 2.
[Section 6 and Appendix D] The adversarial model is underspecified in a way that matters. The text says an adversary can 'arbitrarily corrupt a small fraction η of the trajectories' and cites the strong-contamination model of [23], which normally permits the adversary to depend on the data. The proof, however, requires W_j to be independent of the clean data and of each other—essentially an oblivious adversary. The authors should either provide a proof for the adaptive strong-contamination model or explicitly restrict Theorem 3 to an oblivious adversary and adjust the abstract and Section 6 accordingly. As written, the claimed adversarial guarantee is stronger than what the proof supports.

minor comments (3)

[Appendix B, final paragraph] The condition M ≥ c2(σ̃²/σ²) should read M ≥ c2(σ̃⁴/σ⁴), matching Lemma 2 and Theorem 1; the displayed σ̃²/σ² is a typo.
[Lemma 16, proof] In the displayed Hoeffding expression, the centered sum is written with E[Y_1] but the surrounding text and the subsequent bound use E[Z_1]; this inconsistency is a typo and should be corrected.
[Theorem 3, statement] The requirement K ≥ ⌈c2 log(1/δ) + c3 ηN⌉ is stated with a real argument inside a ceiling, and the feasibility N = MK together with M ≥ c4 d² C_A C_w and the upper bound on η should be stated explicitly; this is a clarity issue about the parameter ranges for which the theorem applies.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the heavy-tailed bounds are derived from the stated fourth-moment assumptions via Chebyshev, Markov, and an external geometric-median lemma; no fitted parameter or load-bearing self-citation appears.

full rationale

The derivation chain is self-contained and does not reduce to its own inputs. The per-bucket OLS error identities (Eq. 7 and Appendix C) are exact algebraic rewritings of $\hat A_j = A + (\sum w_T z_T^\top)(\sum z_T z_T^\top)^{-1}$. The scalar numerator bound (Lemma 1/11) follows from Chebyshev applied to $\mathrm{Var}(x_T w_T)=\sigma^4 g_T$, a consequence of the stated second-moment model. The scalar denominator bound (Lemma 2) and the vector covariance lower bound (Lemma 4/15) are proved from the fourth-moment assumption via Chebyshev and the matrix Markov inequality of Ahlswede and Winter [26], an external source; the resulting dependence on $\tilde\sigma^4/\sigma^4$ and on $C_A$ is an explicit sufficient condition, not a fitted constant. The boosting steps (Lemma 12 and Lemma 16) use Minsker's geometric-median lemma [22], again external, and apply Hoeffding to i.i.d. bucket indicators. No parameter is fitted to a subset of the data, and no target bound is inserted as an assumption. The sub-Gaussian benchmark (3) is quoted from Dean et al. [7], a different group's result, and is used only as a comparison target. The reference list contains no self-citations by the authors, so the self-citation patterns enumerated in the rubric do not arise. The one substantive weakness is in Theorem 3 / Appendix D: the proof centers $Z_j W_j$ by a marginal bound and then applies Hoeffding, but under an adaptive strong-contamination adversary the summands need not be independent and the conditioning argument is not justified. That is a correctness or rigor gap, not a circular reduction; the claimed robustness bound is not equivalent to an input by construction, and Theorems 1 and 2 remain independent of that step.

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

The paper introduces no new entities and fits no free parameters. Its claims rest on the stated heavy-tailed noise model, trajectory independence, zero initial state, and standard concentration and trace inequalities from the cited literature.

assumptions (4)

domain assumption Noise is zero-mean i.i.d. across time with E[w_t w_t^T] = sigma^2 I and E[(w_t(i))^4] = sigma-tilde^4.
Used throughout; the isotropic covariance enables whitening of x_T in Lemma 4 and the scalar fourth-moment computation in Lemma 2.
domain assumption The N trajectories are independent and all start from x_0 = 0.
Independence makes bucket OLS estimates independent for boosting; zero initial condition gives E[x_T x_T^T] = sigma^2 G_T.
standard math Chebyshev, Hoeffding, Markov, and matrix Markov inequalities; von Neumann trace inequality; Rayleigh-Ritz; Minsker's geometric median lemma (Lemma 10).
These are the concentration and linear-algebra tools invoked in Sections B-D and Appendix A.
domain assumption For Theorem 3, the adversary corrupts at most eta N trajectories with eta < 0.5/(c1 d^2 C_A C_w).
The stated corruption bound is part of the theorem statement and enters the proof in Appendix D.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Outlier-Robust Linear System Identification Under Heavy-tailed Noise." pith.science (2026). https://pith.science/paper/NAR5APQV

@misc{pith2026250100421,
  author       = {Pith},
  title        = {Pith review of: Outlier-Robust Linear System Identification Under Heavy-tailed Noise},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NAR5APQV}},
  note         = {Machine review of arXiv:2501.00421}
}

read the original abstract

We consider the problem of estimating the state transition matrix of a linear time-invariant (LTI) system, given access to multiple independent trajectories sampled from the system. Several recent papers have conducted a non-asymptotic analysis of this problem, relying crucially on the assumption that the process noise is either Gaussian or sub-Gaussian, i.e., "light-tailed". In sharp contrast, we work under a significantly weaker noise model, assuming nothing more than the existence of the fourth moment of the noise distribution. For this setting, we provide the first set of results demonstrating that one can obtain sample-complexity bounds for linear system identification that are nearly of the same order as under sub-Gaussian noise. To achieve such results, we develop a novel robust system identification algorithm that relies on constructing multiple weakly-concentrated estimators, and then boosting their performance using suitable tools from high-dimensional robust statistics. Interestingly, our analysis reveals how the kurtosis of the noise distribution, a measure of heavy-tailedness, affects the number of trajectories needed to achieve desired estimation error bounds. Finally, we show that our algorithm and analysis technique can be easily extended to account for scenarios where an adversary can arbitrarily corrupt a small fraction of the collected trajectory data. Our work takes the first steps towards building a robust statistical learning theory for control under non-ideal assumptions on the data-generating process.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Boosting-Enabled Robust System Identification of Partially Observed LTI Systems Under Heavy-Tailed Noise
eess.SY 2025-04 accept novelty 6.0 of 10

A bucket-and-boost estimator using least squares and a geometric median recovers the first T Markov parameters of an LTI system under heavy-tailed noise with O(sqrt(p log(1/delta)/N)) error and O((mT)^2 kappa log(1/de...

Reference graph

Works this paper leans on

34 extracted references · 33 canonical work pages · cited by 1 Pith paper

[7]

On the sam- ple complexity of the linear quadratic regulator

Sarah Dean, Horia Mania, Nikolai Matni, Benjamin Recht, a nd Stephen Tu. On the sam- ple complexity of the linear quadratic regulator. Foundations of Computational Mathematics , 20(4):633–679, 2020

work page 2020
[23]

Robust multivariat e mean estimation: the optimality of trimmed mean

Gabor Lugosi and Shahar Mendelson. Robust multivariat e mean estimation: the optimality of trimmed mean. 2021

work page 2021
[1]

Toward a theoretical foundation of policy optimization for learni ng control policies

Bin Hu, Kaiqing Zhang, Na Li, Mehran Mesbahi, Maryam Fazel , and Tamer Başar. Toward a theoretical foundation of policy optimization for learni ng control policies. Annual Review of Control, Robotics, and Autonomous Systems , 6(1):123–158, 2023

work page 2023
[2]

Statistical learning theory for control: A ﬁnite-sample perspective

Anastasios Tsiamis, Ingvar Ziemann, Nikolai Matni, and George J Pappas. Statistical learning theory for control: A ﬁnite-sample perspective. IEEE Control Systems Magazine , 43(6):67–97, 2023

work page 2023
[3]

High-dimensional probability: An introduction with appli cations in data science, volume 47

Roman Vershynin. High-dimensional probability: An introduction with appli cations in data science, volume 47. Cambridge university press, 2018

work page 2018
[4]

Asymptotic properties of general autor egressive models and strong consis- tency of least-squares estimates of their parameters

TL Lai and CZ Wei. Asymptotic properties of general autor egressive models and strong consis- tency of least-squares estimates of their parameters. Journal of multivariate analysis , 13(1):1– 23, 1983

work page 1983
[5]

Theory for the user

Lennart Ljung et al. Theory for the user. System identiﬁcation , 1987

work page 1987
[6]

Concentration bounds for single parame ter adaptive control

Anders Rantzer. Concentration bounds for single parame ter adaptive control. In 2018 Annual American Control Conference (ACC) , pages 1862–1866. IEEE, 2018

work page 2018

Show all 34 references

[8]

Non-asymptotic identiﬁcation of li near dynamical systems using mul- tiple trajectories

Yang Zheng and Na Li. Non-asymptotic identiﬁcation of li near dynamical systems using mul- tiple trajectories. IEEE Control Systems Letters , 5(5):1693–1698, 2020

2020
[9]

Linear system identiﬁcation under multiplicative noise from multiple tr ajectory data

Yu Xing, Ben Gravell, Xingkang He, Karl Henrik Johansson, and Tyler Summers. Linear system identiﬁcation under multiplicative noise from multiple tr ajectory data. In 2020 American Control Conference (ACC) , pages 5157–5261. IEEE, 2020

2020
[10]

Learning th e dynamics of autonomous linear systems from multiple trajectories

Lei Xin, George Chiu, and Shreyas Sundaram. Learning th e dynamics of autonomous linear systems from multiple trajectories. In 2022 American Control Conference (ACC) , pages 3955–

2022
[11]

Lea rning from many trajectories

Stephen Tu, Roy Frostig, and Mahdi Soltanolkotabi. Lea rning from many trajectories. Journal of Machine Learning Research , 25(216):1–109, 2024

2024
[12]

Learning without mixing: Towards a sharp analysis of linear system id entiﬁcation

Max Simchowitz, Horia Mania, Stephen Tu, Michael I Jord an, and Benjamin Recht. Learning without mixing: Towards a sharp analysis of linear system id entiﬁcation. In Conference On Learning Theory, pages 439–473. PMLR, 2018

2018
[13]

Non-asymptotic identiﬁc ation of lti systems from a single trajectory

Samet Oymak and Necmiye Ozay. Non-asymptotic identiﬁc ation of lti systems from a single trajectory. In 2019 American control conference (ACC) , pages 5655–5661. IEEE, 2019. 12

2019
[14]

Finite-time ide ntiﬁcation of linear systems: Fundamental limits and optimal algorithms

Yassir Jedra and Alexandre Proutiere. Finite-time ide ntiﬁcation of linear systems: Fundamental limits and optimal algorithms. IEEE Transactions on Automatic Control , 68(5):2805–2820, 2022

2022
[15]

Near optimal ﬁnite time identiﬁcation of arbitrary linear dynamical systems

Tuhin Sarkar and Alexander Rakhlin. Near optimal ﬁnite time identiﬁcation of arbitrary linear dynamical systems. In International Conference on Machine Learning , pages 5610–5618. PMLR, 2019

2019
[16]

Finite sample a nalysis of stochastic system iden- tiﬁcation

Anastasios Tsiamis and George J Pappas. Finite sample a nalysis of stochastic system iden- tiﬁcation. In 2019 IEEE 58th Conference on Decision and Control (CDC) , pages 3648–3654. IEEE, 2019

2019
[17]

Lea rning sparse dynamical systems from a single sample trajectory

Salar Fattahi, Nikolai Matni, and Somayeh Sojoudi. Lea rning sparse dynamical systems from a single sample trajectory. In 2019 IEEE 58th Conference on Decision and Control (CDC) , pages 2682–2689. IEEE, 2019

2019
[18]

Finite sample sy stem identiﬁcation: Optimal rates and the role of regularization

Yue Sun, Samet Oymak, and Maryam Fazel. Finite sample sy stem identiﬁcation: Optimal rates and the role of regularization. In Learning for dynamics and control , pages 16–25. PMLR, 2020

2020
[19]

A tutorial on the non-asymptotic theory of system id entiﬁcation

Ingvar Ziemann, Anastasios Tsiamis, Bruce Lee, Yassir J edra, Nikolai Matni, and George J Pappas. A tutorial on the non-asymptotic theory of system id entiﬁcation. In 2023 62nd IEEE Conference on Decision and Control (CDC) , pages 8921–8939. IEEE, 2023

2023
[20]

Finite time identiﬁcation in unstable linear systems

Mohamad Kazem Shirani Faradonbeh, Ambuj Tewari, and Ge orge Michailidis. Finite time identiﬁcation in unstable linear systems. Automatica, 96:342–353, 2018

2018
[21]

The lower tail of random qu adratic forms with applications to ordinary least squares

Roberto Imbuzeiro Oliveira. The lower tail of random qu adratic forms with applications to ordinary least squares. Probability Theory and Related Fields , 166:1175–1194, 2016

2016
[22]

Geometric median and robust estima tion in banach spaces

Stanislav Minsker. Geometric median and robust estima tion in banach spaces. 2015

2015
[24]

Sub-weibull distribu- tions: Generalizing sub-gaussian and sub-exponential pro perties to heavier tailed distributions

Mariia Vladimirova, Stéphane Girard, Hien Nguyen, and Julyan Arbel. Sub-weibull distribu- tions: Generalizing sub-gaussian and sub-exponential pro perties to heavier tailed distributions. Stat, 9(1):e318, 2020

2020
[25]

A tutorial on concentrati on bounds for system identiﬁcation, 2019

Nikolai Matni and Stephen Tu. A tutorial on concentrati on bounds for system identiﬁcation, 2019

2019
[26]

Ahlswede and A

R. Ahlswede and A. Winter. Strong converse for identiﬁc ation via quantum channels, 2001

2001
[27]

Introduction to the non-asymptotic a nalysis of random matrices

Roman Vershynin. Introduction to the non-asymptotic a nalysis of random matrices. arXiv preprint arXiv:1011.3027 , 2010

2010 arXiv
[28]

S ingle trajectory nonparametric learning of nonlinear dynamics

Ingvar M Ziemann, Henrik Sandberg, and Nikolai Matni. S ingle trajectory nonparametric learning of nonlinear dynamics. In conference on Learning Theory , pages 3333–3364. PMLR, 2022

2022
[29]

Lea rning nonlinear dynamical systems from a single trajectory

Dylan Foster, Tuhin Sarkar, and Alexander Rakhlin. Lea rning nonlinear dynamical systems from a single trajectory. In Learning for Dynamics and Control , pages 851–861. PMLR, 2020. 13

2020
[30]

Instance-de pendent ℓinf ty-bounds for policy eval- uation in tabular reinforcement learning

Ashwin Pananjady and Martin J Wainwright. Instance-de pendent ℓinf ty-bounds for policy eval- uation in tabular reinforcement learning. IEEE Transactions on Information Theory, 67(1):566– 585, 2020

2020
[31]

Matrix analysis

Roger A Horn and Charles R Johnson. Matrix analysis . Cambridge university press, 2012. 14 A Preliminary Results and Facts In this section, we provide some of the key results and facts t hat will aid us in the analysis provided in the subsequent appendices. We use the followin...

2012
[32]

≥ λd(A) and λ1(B) ≥ λ2(B) ≥

follows from von Neumann’s trace theorem ( [ 31], Theorem 8.7.6) which tells us that for symmetric positive semi-deﬁnite matrices A and B, we have trace (AB) ≤∑ d i=1 λi(A)λi(B) where λ1(A) ≥ λ2(A) ≥ . . . ≥ λd(A) and λ1(B) ≥ λ2(B) ≥ . . . ≥ λd(B) are ordered eigenval- ues of ...
[33]

follows from the fact that trace (A) =∑ d i=1 λi(A) ≤ dλmax(A) where λmax(A) is the largest eigenvalue of A. Lemma 7. The following hold for any symmetric positive semi-deﬁnite rank 1 matrix A ∈ Rd×d and a symmetric positive semi-deﬁnite matrix B ∈ Rd×d: trace(AB) ≤ trace(A) λ...
[34]

bucket j has no corruptions

follows from the fact that (trace (A))2 = ( ∑ d i=1 λi(A) ) 2 = (λmax(A))2 where the last equality is due to the rank 1 nature of A which allows a single non-zero eigenvalue. Fur- thermore, notice that (λmax(A))2 = trace ( A2) as matrix A2 is also rank 1 with a single non-zero...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [7]

On the sam- ple complexity of the linear quadratic regulator

Sarah Dean, Horia Mania, Nikolai Matni, Benjamin Recht, a nd Stephen Tu. On the sam- ple complexity of the linear quadratic regulator. Foundations of Computational Mathematics , 20(4):633–679, 2020

work page 2020

[2] [23]

Robust multivariat e mean estimation: the optimality of trimmed mean

Gabor Lugosi and Shahar Mendelson. Robust multivariat e mean estimation: the optimality of trimmed mean. 2021

work page 2021

[3] [1]

Toward a theoretical foundation of policy optimization for learni ng control policies

Bin Hu, Kaiqing Zhang, Na Li, Mehran Mesbahi, Maryam Fazel , and Tamer Başar. Toward a theoretical foundation of policy optimization for learni ng control policies. Annual Review of Control, Robotics, and Autonomous Systems , 6(1):123–158, 2023

work page 2023

[4] [2]

Statistical learning theory for control: A ﬁnite-sample perspective

Anastasios Tsiamis, Ingvar Ziemann, Nikolai Matni, and George J Pappas. Statistical learning theory for control: A ﬁnite-sample perspective. IEEE Control Systems Magazine , 43(6):67–97, 2023

work page 2023

[5] [3]

High-dimensional probability: An introduction with appli cations in data science, volume 47

Roman Vershynin. High-dimensional probability: An introduction with appli cations in data science, volume 47. Cambridge university press, 2018

work page 2018

[6] [4]

Asymptotic properties of general autor egressive models and strong consis- tency of least-squares estimates of their parameters

TL Lai and CZ Wei. Asymptotic properties of general autor egressive models and strong consis- tency of least-squares estimates of their parameters. Journal of multivariate analysis , 13(1):1– 23, 1983

work page 1983

[7] [5]

Theory for the user

Lennart Ljung et al. Theory for the user. System identiﬁcation , 1987

work page 1987

[8] [6]

Concentration bounds for single parame ter adaptive control

Anders Rantzer. Concentration bounds for single parame ter adaptive control. In 2018 Annual American Control Conference (ACC) , pages 1862–1866. IEEE, 2018

work page 2018