Pith. sign in

REVIEW 3 major objections 5 minor 48 references

Sparse mean-shift anomalies in multivariate Gaussian data with unknown covariance are exactly detectable when the whitened ell_1 signal crosses sqrt(2)(1+sqrt(1-beta)) sqrt(p log n); the paper gives a plug-in threshold that attains it.

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 ↗

A covariance-whitened ℓ1 threshold √(2(p+δ)log n) detects sparse mean-shift anomalies in p-variate Gaussian samples with asymptotically zero misclassification exactly when ∥L⁻¹a∥₁ exceeds √2(1+√(1−β))√(p log n), with a matching impossibility bound.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection Genuine multivariate extension of the Hamming-loss threshold with a matching lower bound, but the main theorem's estimator conditions are only verified on clean data while the simulations and Section 5 claim otherwise. the 3 major comments →

arxiv 2607.15637 v1 pith:NHDXUO3N submitted 2026-07-17 stat.ME

Asymptotically exact threshold for detecting anomalies in multivariate Gaussian data with application to time series

Marie Tur\v{c}i\v{c}ov\'a , Patr\'icia Martinkov\'a This is my paper
classification stat.ME
keywords anomaly detectionmultivariate Gaussiansparse mean shiftHamming riskexact thresholdHuber covariance estimatorell_1 normtime series residuals
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 reading

The paper studies a sample of n multivariate Gaussian vectors in which a small, unknown fraction are anomalies shifted by a common mean vector, while the covariance matrix is unknown. It proposes a detector that whitens each observation by an estimated Cholesky factor and flags it when the ell_1 norm of the whitened vector exceeds sqrt(2(p+delta)log n), with delta tending to zero slowly. The central claim is a sharp phase transition: expected misclassifications vanish exactly when the whitened signal norm crosses sqrt(2)(1+sqrt(1-beta)) sqrt(p log n), where beta measures anomaly sparsity. A companion impossibility theorem says no estimator can succeed below that level, so the paper identifies both the exact detectability boundary and a procedure that attains it. This matters because the rule removes the usual subjective quantile choices from multivariate outlier detection and applies directly to residuals from time-series models.

Core claim

The paper establishes that, for fixed dimension p and sparsity index beta in (0,1), the anomaly-selection problem in the Gaussian location-slippage model has an asymptotically exact threshold. If liminf of ||L^{-1} a||_1 / sqrt(p log n) is greater than sqrt(2)(1+sqrt(1-beta)), then the plug-in ell_1 detector eta-hat_i = 1{||L-hat^{-1} X_i||_1 > sqrt(2(p+delta) log n)} has Hamming risk tending to zero uniformly over all sparsity patterns with at most n^{1-beta} anomalies. If the same liminf is strictly below that constant, and p is not too large relative to beta, the paper proves that no estimator can drive the Hamming risk to zero over the family H^pm_{n,beta}. Together these theorems locate

What carries the argument

The carrying object is the ell_1 score R(X) = ||L^{-1} X||_1, where L is the Cholesky factor of the covariance Sigma; this reduces each multivariate observation to a scalar that accumulates evidence across components after whitening. The threshold sqrt(2(p+delta) log n) comes from Lemma 2.1, a Gaussian tail bound on the maximum ell_1 norm of n standard normal vectors. The proof of consistency for unknown Sigma relies on decomposition (28) of the covariance estimator into a U-statistic-like average, a deterministic bias, and a centered remainder satisfying assumptions (A1)-(A3), which the paper verifies for the sample covariance from clean data and, at the level of Theorem 2.2, for the Huber-

Load-bearing premise

The theorem requires the estimated covariance matrix to converge to the true one with bias shrinking like 1/n and an 8th-moment remainder of order n^{-4}, yet the paper verifies this only for samples from the clean Gaussian distribution; the Huber estimator used in the simulations and real-data examples is computed from data that contain anomalies, and under contamination at the paper's own sparsity the bias is not O(n^{-1}).

What would settle it

Simulate with fixed p and beta, n large, and a mean shift satisfying ||L^{-1} a||_1 / sqrt(p log n) slightly above sqrt(2)(1+sqrt(1-beta)). Compute the Huber covariance estimate (34)-(35) from the full contaminated sample, run the detector (19)-(20), and track the empirical Hamming risk as n grows. If the bias of the Huber estimator from contamination decays only as n^{-beta} sqrt(log n) rather than n^{-1}, the risk will plateau rather than vanish, contradicting the paper's claim that the contaminated-data pipeline inherits Theorem 2.2.

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

If this is right

  • If the signal condition holds, the plug-in ell_1 detector achieves asymptotically exact anomaly selection uniformly over all sparsity patterns with at most n^{1-beta} anomalies.
  • Because the threshold does not depend on the unknown sparsity beta, the method is adaptive to the anomaly frequency while remaining provably exact at the boundary.
  • The impossibility theorem implies that below the same constant no estimator, however sophisticated, can drive the expected number of misclassifications to zero, so the boundary is minimax for the stated range of p.
  • For p=1 and known variance the theorems reproduce the univariate exact-recovery threshold, so the result is a direct extension of the known variable-selection phase transition.
  • When applied to residuals from ARIMA and vector autoregressive fits, the same threshold flags anomalous days in wearable-device and air-pollution time series without user-chosen quantile levels.

Where Pith is reading between the lines

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

  • The paper leaves implicit that the sharp constant should transfer to other spherically symmetric distributions if the ell_1 norm is replaced by the corresponding Orlicz norm; this is a testable extension, not a claim the paper makes.
  • A natural practical testable corollary is to compute Huber covariance estimates from the data before and after removing flagged anomalies; if the flagged set changes materially, the finite-sample guarantee is fragile even though the asymptotic boundary may stand.
  • Pairing the ell_1 detector with an ell_infinity componentwise check, as the paper informally suggests, would likely lift the boundary for anomalies concentrated in few coordinates; quantifying that two-norm trade-off goes beyond the present theorems.
  • The paper's proof of the Huber covariance assumptions is verified for clean Gaussian data in the appendix, but the simulations and real-data pipeline compute the estimator from contaminated data; a rigorous treatment of that mismatch is needed before the contaminated-data claim can be regarded as established.
Share X Bluesky LinkedIn Reddit HN

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

3 major / 5 minor

Summary. The manuscript proposes a threshold detector (19)–(20) for sparse mean-shift anomalies in multivariate Gaussian data with unknown covariance. Under the sparsity model (1), with anomalies occurring at rate n^{-β}, it defines sets H_{n,β} and H^±_{n,β}. The main positive result, Theorem 2.2, states that if the mean shift satisfies liminf ∥L^{-1}a∥_1/√(p log n) > √2(1+√(1−β)) and the covariance estimator permits decomposition (28) with assumptions (A1)–(A3), then the plug-in ℓ1 detector has Hamming risk tending to zero uniformly over H_{n,β}. Theorem 2.3 provides a matching lower bound for p ≤ ⌊(8/β)(1+√(1−β))^{-4}⌋. The paper verifies (A1)–(A3) for the sample covariance from clean data and for the Huber-type M-estimator (34)–(35), reports simulations comparing the proposed AET method with beta/boxplot/Siotani thresholds, and applies the method to wearable step-count and air-pollution time series.

Significance. If correct, the paper gives a sharp phase-transition constant for fixed-dimensional multivariate Gaussian anomaly selection and shows that a simple ℓ1 plug-in rule saturates it. The conditional theorem is clean, and the proof structure — union bound, Gaussian concentration, Rosenthal inequality, Cholesky perturbation — is standard and largely checkable. The lower bound adapts the Bernoulli-prior Bayes reduction of Butucea et al. (2018), and the manuscript is explicit about the estimator assumptions, which is a strength. The main advertised extension beyond the univariate known-variance setting is robust covariance estimation from data that contain anomalies, but that extension is not actually covered by the hypotheses: the verification in Appendix A.2 is carried out only for clean samples, and the contaminated-data bias is of the wrong order. If that gap is repaired, the result would be a useful contribution; as it stands, the paper supports the conditional statement and the clean-sample covariance case.

major comments (3)
  1. [A.2, Eqs. (34)–(38); Sections 3 and 5] The verification of (A1)–(A3) in Appendix A.2 begins by considering a sample from N_p(μ,Σ), i.e. it covers only the clean case. In the simulations and in the real-data pipeline, Σ̂ is the Huber estimator computed from data generated under model (1), which contains anomalies. Under the signal condition (64), an anomalous observation has Mahalanobis distance M_a = ∥L^{-1}a∥_2^2 of order log n, so the Huber weight u^2(M_a) = r^2/(c M_a) is O(1/log n); multiplying by (a_j+ε_j)(a_k+ε_k) contributes O(1) to the estimating equation. With n^{1-β} anomalies, the deterministic bias in (28) from the anomaly component is Θ(n^{-β}), not O(n^{-1}) as required by (A2). Hence Theorem 2.2 cannot be invoked for the estimator actually used in Sections 3 and 4, and the statement in Section 5 that the Huber-type estimator ‘fulfill[s] these requirements’ in the contaminated case is unsupported. The authors ne
  2. [B, proof of Theorem 2.2, Eqs. (76)–(86)] The proof repeatedly concludes that terms are O(1): see (76), (78), (79), (82), (85), (86). In each displayed estimate the exponent is actually negative (e.g. n^{-1}, n^{-δ/2}, n^{-[(√(1-β)+δ2-δ1)^2]}), so the expressions are o(1). As written, O(1) bounds do not imply that the Hamming risk tends to zero; the final conclusion requires o(1). This is a correctable notational/error-rate issue, but it affects the main proof and must be fixed.
  3. [A.2, around Eqs. (44) and (48)] The verification of (A2)–(A3) for the Huber estimator relies on asymptotic normality of √n(θ̂−θ0) to assert finite-sample moment bounds: ‘√n(σ̂_jk−σ_jk)∼N(0,r) for large n, which yields E∥ΔX∥_F^16 = O(n^{-8})’. Asymptotic distribution does not in general imply convergence of 16th moments; without uniform integrability or a direct Rosenthal-type argument, (A3) is not proved even in the clean case. This is a second gap in the estimator verification.
minor comments (5)
  1. [Throughout] The main text refers to Lemma 1 and Theorems 2.2/2.3, while the appendix uses Lemma 2.1 and equation numbers (64), (87). These numbering systems should be harmonized so that cross-references are unambiguous.
  2. [Section 3, condition (64)] Section 3 refers to ‘condition (64)’, but in the main text the condition is (21); the appendix numbering is used before the appendix is introduced. Please correct the cross-reference.
  3. [Section 3.2] Typo: ‘eight anomaly detection methods ware compared’ should read ‘were compared’.
  4. [Lemma 2.1 vs. detector (19)] Lemma 2.1 is stated with a fixed Δ>0, but the detector uses δ=δ(n)→0 with δ log n→∞. The relationship between these two parameters should be stated explicitly; as written, the lemma’s Δ appears to be a different quantity from the detector’s δ.
  5. [Lemma S1 and proof] The stability condition is written ∥Σ^{-1}∥_2∥Σ−S∥_2 ≤ α < 1 in Lemma S1, but the proof uses the strict inequality ∥Σ^{-1}∥_2∥Σ−S∥_2 < 1. Make the two statements consistent.

Circularity Check

0 steps flagged

No circularity: theorem is conditional on estimator conditions proved in-appendix for clean data; the contaminated-data gap is a correctness issue, not a circular reduction.

full rationale

I walked the derivation chain. Lemma 2.1 is a Gaussian maximum-tail bound proved in Appendix B from a standard chi-square tail approximation and is not derived from the detector's behavior. Theorem 2.2 is explicitly conditional: if the covariance estimator admits decomposition (28) with (A1)-(A3), then the plug-in ell-1 detector (19)-(20) is asymptotically exact; its proof in Appendix B proceeds directly from those hypotheses. Theorem 2.3 is a minimax lower bound using the Bernoulli-prior Bayes reduction of [8], cited and acknowledged in Remark 2; this is external inherited technique, not a self-citation chain. The threshold itself contains no fitted parameter, so there is no fitted-input-called-prediction issue. The self-citations [36]-[38] are only literature-extension remarks, not load-bearing. The substantive caveat is a hypothesis-verification gap, not circularity: Appendix A.2 opens with 'Consider a sample X1,...,Xn from N_p(mu,Sigma)' and verifies (A1)-(A3) only in the clean case, while Section 5 states that 'the Huber-type estimator (based on data contaminated by anomalies)' was shown to achieve zero expected misclassifications. If the anomaly contamination makes the Huber bias of order n^{-beta} rather than n^{-1}, Assumption (A2) fails for the pipeline as run, so Theorem 2.2 cannot be invoked as proven. That is a correctness risk about whether the theorem's hypotheses hold, not a case where the claimed prediction is equivalent by construction to the paper's inputs.

Axiom & Free-Parameter Ledger

4 free parameters · 8 axioms · 0 invented entities

The central claim rests on: (i) the Gaussian maximum tail, an external standard result with the correct constant; (ii) the estimator decomposition (A1)–(A3), whose verification is the least rigorous part of the paper and is provided only for clean data while the deployed procedure uses contaminated data; (iii) the lower-bound machinery inherited from [6]/[8]. The threshold contains one hand-chosen slack δ; β enters only through the stated guarantees. No new physical/statistical entities are postulated: R(X)=∥L⁻¹X∥₁ is an estimator, and (28) is a condition, not an entity with independent falsifiable handles.

free parameters (4)
  • δ(n) — threshold slack = δ = 1/√log n in simulations and real-data analyses
    Appears in the threshold (19) √(2(p+δ)log n); constrained only by δ→0 and δ log n→∞ (condition (20)); chosen by hand in practice, not fitted to data.
  • β — sparsity index = β=0.4 (Fig. 2), β=0.5 (Tables 1–2); unknown to the method
    Defines the anomaly budget n^{1−β} and the phase-transition constant √2(1+√(1−β)) in Theorems 2.2–2.3. The detector is adaptive to β, but the guarantees and the lower-bound p-restriction depend on it.
  • κ, r, c — Huber estimator settings = κ=0.05, r = 95% quantile of χ²_p, c = ((−2 f_{χ²_p}(r) + κ) r/p + 1 − κ)
    Standard defaults of robmed::cov.Huber used in §3–§4; chosen by hand, not fitted to the anomaly problem. They determine the covariance estimator used in the pipeline, not the detector.
  • c₀, c₁ — constants in H±_{n,β} = unspecified (0 < c₀ < 1 < c₁ < ∞)
    Bound the anomaly count in the lower-bound set H±_{n,β}; arbitrary constants that do not affect rates.
axioms (8)
  • domain assumption Model (1): X_i = η_i a + ε_i, ε_i ~ N_p(0,Σ) i.i.d., clean mean exactly zero, anomalies are additive mean shifts with η_i fixed indicators
    Stated in §2; all theorems and the detector live inside this model; real-data residuals are asserted to be approximately Gaussian and centered, which is only checked visually via QQ-plots and ACF plots.
  • domain assumption Covariance estimator decomposition (28) with (A1)–(A3): E|g_jk(X₁)−σ_jk|⁸ < ∞, deterministic bias b_jk = O(n⁻¹), centered remainder with E|R_jk|⁸ = O(n⁻⁴)
    Hypothesis of Theorem 2.2; verified in A.2 for sample covariance and Huber estimator under the clean normal model only. The paper's own §5 extends the claim to contaminated data without proof.
  • domain assumption Stability condition ∥Σ⁻¹∥₂∥Σ−S∥₂ ≤ α < 1 for large n (Lemma S1)
    Needed so the inverse Cholesky factor L̂⁻¹ is locally Lipschitz in S; invoked throughout the proof of Theorem 2.2 (I_a¹, I_a², Eq. (83)). The paper notes it held in all simulation runs.
  • standard math Gaussian concentration inequality (Boucheron–Lugosi–Massart, Thm 5.6) for the √p-Lipschitz ℓ1 norm
    Used in the false-positive and false-negative tail bounds (77)–(78) and (84).
  • standard math Rosenthal inequality (Petrov, Thm 2.9) applied at r=8
    Controls the 8th moments of the empirical part of the covariance estimator, Eqs. (73)–(74).
  • standard math χ² tail approximation (Abramowitz–Stegun 26.4.12) used in Lemma 2.1
    Yields P(max_j ∥Z_j∥₁ ≥ √(2(p+Δ)log n)) = O(n^{−Δ/p}(log n)^{p/2−1}) → 0.
  • standard math Bernoulli(n^{−β}) prior and π(B_{n,β}) ≥ 1 − O(n⁻¹), cited from [8, Lemma 2], plus reduction of minimax risk to a single-test Bayes risk
    The lower bound Theorem 2.3 is inherited from the [6]/[8] reduction; the paper restates the argument but relies on the prior-concentration lemma from [8] without reproducing it.
  • ad hoc to paper First-order perturbation expansion (Σ+Δ)⁻¹ ≈ Σ⁻¹ − Σ⁻¹ΔΣ⁻¹ without tracked remainder, and asymptotic normality of √n(θ̂−θ₀) treated as an exact 16th-moment bound E∥Δ∥¹⁶ = O(n⁻⁸)
    Appendix A.2 uses these to claim E|R_jk|⁸ = O(n⁻⁴) for the Huber estimator; no remainder control or uniform-integrability argument is supplied, so the 8th-moment verification is not rigorous as written.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Asymptotically exact threshold for detecting anomalies in multivariate Gaussian data with application to time series." pith.science (2026). https://pith.science/paper/NHDXUO3N

@misc{pith2026260715637,
  author       = {Pith},
  title        = {Pith review of: Asymptotically exact threshold for detecting anomalies in multivariate Gaussian data with application to time series},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NHDXUO3N}},
  note         = {Machine review of arXiv:2607.15637}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In this paper, we propose a new thresholding technique for detecting anomalies in multivariate normal random samples, under the assumption that anomalous observations are sparse and differ from the rest of the data in their mean. The mean vector of the non-anomalous data is assumed to be zero, while the covariance matrix is unknown. We derive conditions on the mean shift of the anomalous observations, as well as on the covariance matrix and its estimator, under which the proposed procedure achieves asymptotically exact detection, meaning that the expected number of misclassified observations converges to zero as the sample size increases. In addition, we establish conditions under which exact anomaly detection is impossible for any procedure. The performance of the proposed method is illustrated through an extensive simulation study and compared with other widely used anomaly detection methods. Real-data analyses involving wearable activity measurements and air pollution time series provide an assessment of its performance in real-world settings.

Figures

Figures reproduced from arXiv: 2607.15637 by Marie Tur\v{c}i\v{c}ov\'a, Patr\'icia Martinkov\'a.

Figure 2
Figure 2. Figure 2: In [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 1
Figure 1. Figure 1: Components of the expected value an = (an1, an2, an3, an4) ⊤ of the anomalies depending on the sample size n. in the structure of the mean shift of the anomalies. In the first scenario, the anomalies are caused by a large magnitude of one component (the mean shift is set to the value a1 below). In the second scenario, the anomalies are caused by mild deviations in all components (the mean shift equals a2 b… view at source ↗
Figure 2
Figure 2. Figure 2: Average number of misclassified observations (false positives + false negatives), which estimates [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Detection in a Gaussian sample {Xj = (Xj1, . . . , Xj4) ⊤, j = 1, . . . , 800}, with mean a1 and covariance specified in (26). Compared multivariate methods: BD: threshold (7) based on beta distribution, BPR: boxplot rule (8), AET: asymptotically exact threshold (19)–(20), ST: Siotani’s threshold (9), and univariate methods: AET1: asymptotically exact threshold (19)–(20) with p = 1, GT: Grubbs test (4), 3s… view at source ↗
Figure 4
Figure 4. Figure 4: Detection in a Gaussian sample {Xj = (Xj1, . . . , Xj4) ⊤, j = 1, . . . , 800}, with mean a2 and covariance specified in (26). Compared multivariate methods: BD: threshold (7) based on beta distribution, BPR: boxplot rule (8), AET: asymptotically exact threshold (19)–(20), ST: Siotani’s threshold (9) and univariate methods: AET1: asymptotically exact threshold (19)–(20) with p = 1, GT: Grubbs test (4), 3s:… view at source ↗
Figure 5
Figure 5. Figure 5: Daily step counts with detected anomalies for the participant with ID 999. Time period 409 days [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Daily step counts with detected anomalies for the participant with ID 1429. Time period 412 days [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Daily step counts with detected anomalies for the participant with ID 2110. Time period 373 days [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Air pollutant concentrations with detected anomalous days and the official Air Quality Index in [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Participant with ID 999. QQ-plot and the autocorrelation function for model residuals. Time [PITH_FULL_IMAGE:figures/full_fig_p034_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Participant with ID 1429. QQ-plot and the autocorrelation function for model residuals. Time [PITH_FULL_IMAGE:figures/full_fig_p034_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Participant with ID 2110. QQ-plot and the autocorrelation function for model residuals. Time [PITH_FULL_IMAGE:figures/full_fig_p034_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: QQ-plot and autocorrelation function of the residuals from the VAR(4) model for pollutants NO [PITH_FULL_IMAGE:figures/full_fig_p035_12.png] view at source ↗

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

48 extracted references

  1. [1]

    Stegun.Handbook of Mathematical Tables with Formulas, Graphs, and Mathematical Tables

    Milton Abramowitz and Irene A. Stegun.Handbook of Mathematical Tables with Formulas, Graphs, and Mathematical Tables. Number 55 in Applied Mathematics Series. National Bureau of Standards, Washington, DC, USA, 10th edition, 1972

    1972

  2. [2]

    John Wiley & Sons, New Jersey, 3rd edition, 2003

    Theodore Wilbur Anderson.An Introduction to Multivariate Statistical Analysis. John Wiley & Sons, New Jersey, 3rd edition, 2003

    2003

  3. [3]

    John Wiley & Sons, Chichester, 3rd edition, 1998

    Vic Barnett and Toby Lewis.Outliers in Statistical Data. John Wiley & Sons, Chichester, 3rd edition, 1998

    1998

  4. [4]

    Automatic outlier detection for time series: an application to sensor data.Knowledge and Information Systems, 11(2):137–154, 2007

    Sabyasachi Basu and Martin Meckesheimer. Automatic outlier detection for time series: an application to sensor data.Knowledge and Information Systems, 11(2):137–154, 2007

    2007

  5. [5]

    Oxford University Press, Oxford, 1st edition, 2013

    St´ ephane Boucheron, G´ abor Lugosi, and Pascal Massart.Concentration Inequalities: A Nonasymptotic Theory of Independence. Oxford University Press, Oxford, 1st edition, 2013

    2013

  6. [6]

    Variable selection with Hamming loss.The Annals of Statistics, 46(5):1837–1875, 2018

    Cristina Butucea, Mohamed Ndaoud, Natalia Stepanova, and Alexandre Tsybakov. Variable selection with Hamming loss.The Annals of Statistics, 46(5):1837–1875, 2018. 18

    2018

  7. [7]

    Anomaly detection: A survey.ACM Computing Surveys, 41(3):15, 2009

    Varun Chandola, Arindam Banerjee, and Vipin Kumar. Anomaly detection: A survey.ACM Computing Surveys, 41(3):15, 2009

    2009

  8. [8]

    Optimal component selection in high dimension

    Xin Cui. Optimal component selection in high dimension. Master’s thesis, Carleton University, Ottawa, Canada, 2014

    2014

  9. [9]

    Index kvality ovzduˇ s ´ ı, 2026

    Czech Hydrometeorological Institute. Index kvality ovzduˇ s ´ ı, 2026. Accessed: 2026-04-20

    2026

  10. [10]

    The temporal dynamics of the association between daily physical activity and life satisfaction.Annals of Behavioral Medicine, 59(1):1–14, 2025

    Steriani Elavsky, Marek Brabec, Marek Maly, Lenka Knapova, Barbora Kastovska, Michal Sebera, Marcela Ely, Vera K Jandackova, Jan Keller, and Misha Pavel. The temporal dynamics of the association between daily physical activity and life satisfaction.Annals of Behavioral Medicine, 59(1):1–14, 2025

    2025

  11. [11]

    European air quality index, 2026

    European Environment Agency. European air quality index, 2026. Accessed: 2026-04-20

    2026

  12. [12]

    Kettenring

    Ramanathan Gnanadesikan and John R. Kettenring. Robust estimates, residuals, and outlier detection with multiresponse data.Biometrics, 28(1):81–124, 1972

    1972

  13. [13]

    Eea’s revision of the european air quality index bands

    Alberto Gonz´ alez Ortiz, Joana Soares, Jaume Targa, Maria Colina, Laia Azc´ on, Lorena Banyuls, Myriam Tobollik, Silvia Monge, and Ragnhildur Finnbjornsdottir. Eea’s revision of the european air quality index bands. ETC HE Report 2024/17, European Topic Centre on Human Health and the Environment, 2025

    2024

  14. [14]

    Frank E. Grubbs. Procedures for detecting outlying observations in samples.Technometrics, 11(1):1–21, 1969

    1969

  15. [15]

    Hawkins.Identification of Outliers

    Douglas M. Hawkins.Identification of Outliers. Chapman & Hall, 1980

    1980

  16. [16]

    Higham.Functions of Matrices: Theory and Computation

    Nicholas J. Higham.Functions of Matrices: Theory and Computation. Society for Industrial and Applied Mathematics, Philadelphia, 2008

    2008

  17. [17]

    Hill and Barbara S

    David J. Hill and Barbara S. Minsker. Anomaly detection in streaming environmental sensor data: A data-driven modeling approach.Environmental Modelling & Software, 25(9):1014–1022, 2010

    2010

  18. [18]

    Horn and Charles R

    Roger A. Horn and Charles R. Johnson.Matrix Analysis. Cambridge University Press, United States of America, 2nd edition, 2013

    2013

  19. [19]

    Multidimensional two-component gaussian mixtures detection.Annales de l’Institut Henri Poincar´ e - Probabilit´ es et Statistiques, 54(2):842—-865, 2018

    B´ eatrice Laurent, Cl´ ement Marteau, and Cathy Maugis-Rabusseau. Multidimensional two-component gaussian mixtures detection.Annales de l’Institut Henri Poincar´ e - Probabilit´ es et Statistiques, 54(2):842—-865, 2018

    2018

  20. [20]

    Informal identification of outliers in medical data

    Jorma Laurikkala, Martti Juhola, and Erna Kentala. Informal identification of outliers in medical data. InProceedings of the Fifth International Workshop on Intelligent Data Analysis in Medicine and Pharmacology, 2000

    2000

  21. [21]

    Online learning and sequential anomaly detection in trajecto- ries.IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(6):1158–1173, 2014

    Rikard Laxhammar and G¨ oran Falkman. Online learning and sequential anomaly detection in trajecto- ries.IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(6):1158–1173, 2014

    2014

  22. [22]

    Lehmann and Joseph P

    Erich L. Lehmann and Joseph P. Romano.Testing Statistical Hypotheses. Springer, New York, 3rd edition, 2005

    2005

  23. [23]

    Detection of outlying data in bioavailability/bioequivalence studies

    Jen-Pei Liu and Chung-Sing Weng. Detection of outlying data in bioavailability/bioequivalence studies. Statistics in medicine, 10:1375–1389, 1991

    1991

  24. [24]

    Detecting and tracking regional outliers in meteorological data.Information Sciences, 177:1609–1632, 2007

    Chang-Tien Lu, Yufeng Kou, Jiang Zhao, and Li Chen. Detecting and tracking regional outliers in meteorological data.Information Sciences, 177:1609–1632, 2007

    2007

  25. [25]

    Ricardo A. Maronna. Robust M-estimators of multivariate location and scatter.The Annals of Statistics, 4(1):51–67, 1976

    1976

  26. [26]

    Petrov.Limit Theorems of Probability Theory: Sequences of Independent Random Variables

    Valentin V. Petrov.Limit Theorems of Probability Theory: Sequences of Independent Random Variables. Clarendon Press, Oxford, 1995

    1995

  27. [27]

    S.W. Roberts. Control chart tests based on geometric moving averages.Technometrics, 1(3):239–250, 1959. 19

    1959

  28. [28]

    Rousseeuw and Annick M

    Peter J. Rousseeuw and Annick M. Leroy.Robust Regression and Outlier Detection. John Wiley & Sons, New York, 1987

    1987

  29. [29]

    Schreuder, and Eva Ceulemans

    Evelien Schat, Marieke J. Schreuder, and Eva Ceulemans. Statistical process control for real-time monitoring in clinical psychology: State of the art and future research agenda.Neuroscience Applied, 5, 2026

    2026

  30. [30]

    Anomaly detection in time series: A comprehensive evaluation.Proceedings of the VLDB Endowment, 15(9):1779–1797, 2022

    Sebastian Schmidl, Phillip Wenig, and Thorsten Papenbrock. Anomaly detection in time series: A comprehensive evaluation.Proceedings of the VLDB Endowment, 15(9):1779–1797, 2022

    2022

  31. [31]

    Shewhart.Economic Control of Quality of Manufactured Product

    Walter A. Shewhart.Economic Control of Quality of Manufactured Product. D. Van Nostrand Company, Inc., New York, 1931

    1931

  32. [32]

    Anomaly detection in streams with extreme value theory.Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1067–1075, 2017

    Alban Siffer, Pierre-Alain Fouque, Alexandre Termier, and Christine Largouet. Anomaly detection in streams with extreme value theory.Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1067–1075, 2017

    2017

  33. [33]

    The extreme value of the generalized distances of the individual points in the multivariate normal sample.Annals of the Institute of Statistical Mathematics, 10:183–208, 1959

    Minoru Siotani. The extreme value of the generalized distances of the individual points in the multivariate normal sample.Annals of the Institute of Statistical Mathematics, 10:183–208, 1959

    1959

  34. [34]

    Rejecting outliers by maximum normed residual.The Annals of Mathematical Statistics, 42(1):35–45, 1971

    Wilhelmine Stefansky. Rejecting outliers by maximum normed residual.The Annals of Mathematical Statistics, 42(1):35–45, 1971

    1971

  35. [35]

    Rejecting outliers in factorial designs.Technometrics, 14(2):469–479, 1972

    Wilhelmine Stefansky. Rejecting outliers in factorial designs.Technometrics, 14(2):469–479, 1972

    1972

  36. [36]

    Adaptive exact recovery in sparse nonparametric models

    Natalia Stepanova and Marie Turcicova. Adaptive exact recovery in sparse nonparametric models. Statistical Inference for Stochastic Processes, 28:1–26, 2025

    2025

  37. [37]

    Exact variable selection in sparse nonparametric models.Elec- tronic Journal of Statistics, 19:2001–2032, 2025

    Natalia Stepanova and Marie Turcicova. Exact variable selection in sparse nonparametric models.Elec- tronic Journal of Statistics, 19:2001–2032, 2025

    2001

  38. [38]

    Adaptive almost full recovery in sparse nonpara- metric models

    Natalia Stepanova, Marie Turcicova, and Xiang Zhao. Adaptive almost full recovery in sparse nonpara- metric models. Preprint on arXiv:2512.10488, 2025

    arXiv 2025

  39. [39]

    Stewart and Ji-guang Sun.Matrix Perturbation Theory

    Gilbert W. Stewart and Ji-guang Sun.Matrix Perturbation Theory. Academic Press, San Diego, 1st edition, 1990

    1990

  40. [40]

    Perturbation bounds for the cholesky and QR factorizations.BIT Numerical Mathematics, 31:341–352, 1991

    Ji-guang Sun. Perturbation bounds for the cholesky and QR factorizations.BIT Numerical Mathematics, 31:341–352, 1991

    1991

  41. [41]

    Tsay, Daniel Pe˜ na, and Alan E

    Ruey S. Tsay, Daniel Pe˜ na, and Alan E. Pankratz. Outliers in multivariate time series.Biometrika, 87(4):789–804, 2000

    2000

  42. [42]

    David E. Tyler. Robustness and efficiency properties of scatter matrices.Biometrika, 70(2):411–420, 1983

    1983

  43. [43]

    Cambridge University Press, 1st edition, 1998

    Adrianus Willem van der Vaart.Asymptotic Statistics. Cambridge University Press, 1st edition, 1998

    1998

  44. [44]

    Cambridge University Press, 2018

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

    2018

  45. [45]

    Detection and feature selection in sparse mixture models.The Annals of Statistics, 45(5):1920–1950, 2017

    Nicolas Verzelen and Ery Arias-Castro. Detection and feature selection in sparse mixture models.The Annals of Statistics, 45(5):1920–1950, 2017

    1920

  46. [46]

    Improving time series anomaly detection based on exponentially weighted moving average (ewma) of season-trend model residuals

    Zeng-Guang Zhou and Ping Tang. Improving time series anomaly detection based on exponentially weighted moving average (ewma) of season-trend model residuals. In2016 IEEE International Geo- science and Remote Sensing Symposium (IGARSS), pages 3414–3417, 2016

    2016

  47. [47]

    nX i=1 u2(M ˆΣ(Xi))−u 2(MΣ(Xi)) (Xij −µ j)(Xik −µ k) # (37) bjk (n) = 1 n E

    Jiyun Zu and Ke-Hai Yuan. Local influence and robust procedures for mediation analysis.Multivariate Behavioral Research, 45(1):1–44, 2010. 20 A Decomposition (12) of a covariance estimator Consider a sampleX 1, . . . ,Xn fromN p(µ,Σ), where Σ = (σ jk )p j,k=1 is unknown. In Section 2.3, we assume the covariance estimatorS= (s jk )p j,k=1 to allow for the ...

    2010

  48. [48]

    (81) By applying the same arguments as forI a 1 withY i =X i −ain place ofX i and using the bound Pn i=1 ηi ≤n, we get a result analogous to (76), I a 2 =O(1).(82) Note that for all sufficiently largen, we can write p (2 +δ) lognin the form √2 logn(1 +δ 1) with some δ1 =O(1). Furthermore, assumption (64) implies that for some positiveδ 2 =O(1) and all suf...

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.