Robust Analysis for Resilient AI System

Lulu Kang; Ran Jin; Yu Wang

arxiv: 2509.06172 · v1 · submitted 2025-09-07 · 📊 stat.AP · cs.LG

Robust Analysis for Resilient AI System

Yu Wang , Ran Jin , Lulu Kang This is my paper

Pith reviewed 2026-05-18 18:52 UTC · model grok-4.3

classification 📊 stat.AP cs.LG

keywords DPD-Lassorobust regressiondensity power divergencelasso regularizationoutlier contaminationindustrial AIresilient systemsmanufacturing data

0 comments

The pith

DPD-Lasso integrates density power divergence with lasso regularization to reliably analyze outlier-contaminated data from industrial AI resilience experiments.

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

The paper introduces DPD-Lasso as a robust regression method that combines density power divergence and lasso regularization to handle severe data outliers generated by operational hazards in manufacturing industrial internet systems. It develops an efficient iterative algorithm to solve the optimization problem and overcome prior computational bottlenecks. When applied to an aerosol jet printing testbed, the approach shows stable performance on both clean and contaminated data while accurately quantifying hazard impacts. This work positions robust regression as a necessary tool for developing and validating resilient industrial AI systems.

Core claim

DPD-Lasso provides reliable, stable performance on both clean and outlier-contaminated data from AI resilience experiments, accurately quantifying hazard impacts by integrating density power divergence with lasso regularization and solving the resulting optimization through a new iterative algorithm.

What carries the argument

DPD-Lasso, the robust regression estimator formed by fusing density power divergence for outlier resistance with lasso regularization for variable selection, solved by an iterative algorithm.

If this is right

DPD-Lasso enables accurate quantification of hazard impacts even when operational hazards produce severe outliers in manufacturing data.
The method supports validation of resilient industrial AI systems by maintaining reliable analysis on both clean and contaminated datasets.
The iterative solver makes density power divergence methods computationally practical for lasso-regularized problems.
Robust regression becomes essential for testing AI performance in real manufacturing environments subject to data contamination.

Where Pith is reading between the lines

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

The same divergence-plus-regularization structure could be tested on sensor streams from other outlier-prone domains such as autonomous vehicles or energy grids.
Explicit convergence bounds or error guarantees for the iterative solver would be a natural next step to support wider deployment.
The results suggest that classical statistical estimators may require systematic robustness upgrades before use in deployed industrial AI pipelines.

Load-bearing premise

The iterative algorithm solves the DPD-Lasso optimization accurately and stably without detailed convergence analysis or error bounds.

What would settle it

Direct comparison of DPD-Lasso hazard estimates against known ground-truth values in the aerosol jet printing testbed under controlled levels of outlier contamination.

Figures

Figures reproduced from arXiv: 2509.06172 by Lulu Kang, Ran Jin, Yu Wang.

**Figure 1.** Figure 1: Box plots of RMSPE, L2 Error of β, and variable selection error γ of 100 simulations of four different methods. (a) Comparison of RMSPE for Inference Time Prediction. (b) Comparison of RMSPE for F1 Score Prediction [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗

**Figure 2.** Figure 2: Comparison of RMSPE for Different Prediction Objects. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Figure 3: Panel of 10 contour plots of least square loss, [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

read the original abstract

Operational hazards in Manufacturing Industrial Internet (MII) systems generate severe data outliers that cripple traditional statistical analysis. This paper proposes a novel robust regression method, DPD-Lasso, which integrates Density Power Divergence with Lasso regularization to analyze contaminated data from AI resilience experiments. We develop an efficient iterative algorithm to overcome previous computational bottlenecks. Applied to an MII testbed for Aerosol Jet Printing, DPD-Lasso provides reliable, stable performance on both clean and outlier-contaminated data, accurately quantifying hazard impacts. This work establishes robust regression as an essential tool for developing and validating resilient industrial AI systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

DPD-Lasso combines density power divergence with Lasso for outlier handling in manufacturing AI data, but the iterative solver lacks the convergence or error analysis needed to back the stability claims.

read the letter

The paper introduces DPD-Lasso as a way to run regression on contaminated industrial data from MII systems. It pairs density power divergence with Lasso regularization and tests the result on an Aerosol Jet Printing setup, claiming stable estimates of hazard effects even when outliers are present. That combination for this specific application is the main new piece, and the testbed example shows it can produce usable output on both clean and dirty versions of the data. The practical framing around resilient AI in manufacturing is a plus; it ties the method to a real workflow instead of leaving it abstract. The iterative solver is presented as fixing earlier computational problems, which is a reasonable engineering step if it works. The soft spot is exactly where the stress-test note points: no convergence rates, fixed-point checks, or bounds on approximation error appear in the description. DPD estimators can be sensitive to how precisely the objective is solved, especially on heavy-tailed data, so the reported stability on contaminated runs could partly reflect solver behavior rather than the estimator itself. Without those details the central reliability claim stays harder to evaluate. This is aimed at applied statisticians and engineers who need robust tools for industrial sensor or process data. Readers working on similar manufacturing AI problems could pick up the case study and the basic idea. It is worth sending to peer review because the domain application is concrete and the method is a straightforward extension that might help practitioners, even though the solver analysis needs to be added before the claims land solidly.

Referee Report

1 major / 1 minor

Summary. The manuscript proposes DPD-Lasso, a robust regression method that integrates Density Power Divergence with Lasso regularization to analyze outlier-contaminated data from Manufacturing Industrial Internet (MII) systems. The authors develop an efficient iterative algorithm to solve the resulting optimization problem and apply the method to an Aerosol Jet Printing testbed, claiming that it delivers reliable, stable performance on both clean and contaminated data while accurately quantifying hazard impacts.

Significance. If the central claims hold after addressing the gaps below, the work would offer a practically useful extension of robust regression techniques to industrial AI resilience problems, where operational hazards routinely produce severe outliers that defeat standard methods. The application to a real MII testbed is a strength, but the absence of quantitative validation, baseline comparisons, and theoretical guarantees currently limits the assessed significance.

major comments (1)

The description of the iterative algorithm developed to solve the DPD-Lasso objective provides no convergence rates, fixed-point analysis, or bounds on approximation error in terms of the density power parameter, regularization strength, or contamination fraction. This is load-bearing for the central claim of reliable performance on contaminated data, because DPD-based estimators are known to be sensitive to solver precision on heavy-tailed observations; without such guarantees, observed stability on the testbed could be an algorithmic artifact rather than a property of the estimator.

minor comments (1)

The abstract states that DPD-Lasso 'provides reliable, stable performance' and 'accurately quantifying hazard impacts' yet contains no numerical results, comparison metrics, or error measures; adding at least one quantitative highlight would improve clarity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We appreciate the emphasis on theoretical guarantees for the iterative algorithm and address this point directly below. We will revise the manuscript to incorporate the requested analysis.

read point-by-point responses

Referee: The description of the iterative algorithm developed to solve the DPD-Lasso objective provides no convergence rates, fixed-point analysis, or bounds on approximation error in terms of the density power parameter, regularization strength, or contamination fraction. This is load-bearing for the central claim of reliable performance on contaminated data, because DPD-based estimators are known to be sensitive to solver precision on heavy-tailed observations; without such guarantees, observed stability on the testbed could be an algorithmic artifact rather than a property of the estimator.

Authors: We agree that formal convergence analysis would strengthen the central claims. The current manuscript presents the iterative algorithm (an alternating weighted Lasso solver derived from the DPD objective) and demonstrates its practical performance on the Aerosol Jet Printing data, but does not include rates or error bounds. In the revised version we will add a dedicated subsection deriving linear convergence to a stationary point under the restricted eigenvalue condition on the design matrix, with explicit dependence on the density power parameter, regularization strength, and contamination fraction. We will also supply numerical verification of convergence speed and error control on synthetic data calibrated to the MII testbed characteristics. These additions will show that the observed stability arises from the estimator rather than solver artifacts. revision: yes

Circularity Check

0 steps flagged

No significant circularity in DPD-Lasso derivation or claims

full rationale

The paper proposes DPD-Lasso by combining Density Power Divergence with Lasso regularization as a new robust regression approach for outlier-contaminated manufacturing data, then describes an iterative solver and reports its empirical performance on an Aerosol Jet Printing testbed for both clean and contaminated cases. No derivation step equates a claimed prediction or result to its own inputs by construction, no fitted parameter is relabeled as an independent prediction, and no load-bearing uniqueness or ansatz is imported solely via self-citation. The central claims rest on the explicit construction of the estimator and its observed stability rather than tautological reduction or renaming of prior patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Limited details available from abstract only; relies on standard statistical assumptions for divergence-based robust regression and optimization convergence.

axioms (1)

domain assumption Standard assumptions for density power divergence and Lasso regularization hold in the presence of data contamination from operational hazards.
Invoked implicitly to justify the method's applicability to MII systems.

pith-pipeline@v0.9.0 · 5614 in / 1084 out tokens · 37429 ms · 2026-05-18T18:52:17.668239+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Q_α(β, σ²) = -log( (1/n) Σ exp(-α (y_i - x_iᵀβ)² / (2σ²)) ) + λ‖β‖₁ (Eq. 6); iterative weighted Lasso via first-order Taylor (Eq. 7, Algorithm 1)
IndisputableMonolith/Foundation/AlphaCoordinateFixation.lean J_uniquely_calibrated_via_higher_derivative unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

When α=1 this becomes the L2E robust regression problem

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

[1]

Fair: Facilitating artifi- cial intelligence resilience in manufacturing industrial internet,

Y . Zeng, I. Lourentzou, X. Deng, and R. Jin, “Fair: Facilitating artifi- cial intelligence resilience in manufacturing industrial internet,”arXiv preprint arXiv:2503.01086, 2025

work page arXiv 2025
[2]

Robust estimation of a location parameter,

P. J. Huber, “Robust estimation of a location parameter,”The Annals of Mathematical Statistics, vol. 35, no. 1, pp. 73–101, 1964

work page 1964
[3]

Asymptotic normality of r-estimates in the linear model,

S. Heiler and R. Willers, “Asymptotic normality of r-estimates in the linear model,”Statistics, vol. 19, no. 2, pp. 173–184, 1988. [Online]. Available: https://doi.org/10.1080/02331888808802084

work page doi:10.1080/02331888808802084 1988
[4]

Robust regression by means of s- estimators,

P. Rousseeuw and V . Yohai, “Robust regression by means of s- estimators,” inRobust and Nonlinear Time Series Analysis, J. Franke, W. H¨ardle, and D. Martin, Eds. New York, NY: Springer US, 1984, pp. 256–272

work page 1984
[5]

Robust linear model selection based on least angle regression,

J. A. Khan, S. V . Aelst, and R. H. Zamar, “Robust linear model selection based on least angle regression,”Journal of the American Statistical Association, vol. 102, no. 480, pp. 1289–1299, 2007

work page 2007
[6]

Robust groupwise least angle regression,

A. Alfons, C. Croux, and S. Gelper, “Robust groupwise least angle regression,”Computational Statistics & Data Analysis, vol. 93, pp. 421–435, 2016. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S0167947315000468

work page 2016
[7]

Robust lasso regression using tukey’s biweight criterion,

L. Chang, S. Roberts, and A. Welsh, “Robust lasso regression using tukey’s biweight criterion,”Technometrics, vol. 60, no. 1, pp. 36–47, 2018

work page 2018
[8]

Statistical consistency and asymptotic normality for high-dimensional robustM-estimators,

P.-L. Loh, “Statistical consistency and asymptotic normality for high-dimensional robustM-estimators,”The Annals of Statistics, vol. 45, no. 2, pp. 866–896, 2017. [Online]. Available: https: //doi.org/10.1214/16-AOS1471

work page doi:10.1214/16-aos1471 2017
[9]

Robust and efficient estimation by minimising a density power divergence,

A. Basu, I. R. Harris, N. L. Hjort, and M. Jones, “Robust and efficient estimation by minimising a density power divergence,”Biometrika, vol. 85, no. 3, pp. 549–559, 1998

work page 1998
[10]

Robust parameter estimation with a small bias against heavy contamination,

H. Fujisawa and S. Eguchi, “Robust parameter estimation with a small bias against heavy contamination,”Journal of Multivariate Analysis, vol. 99, no. 9, pp. 2053–2081, 2008

work page 2053
[11]

The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming,

L. Bregman, “The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming,”USSR Computational Mathematics and Mathematical Physics, vol. 7, no. 3, pp. 200–217, 1967. [Online]. Available: https://www.sciencedirect.com/science/article/pii/0041555367900407

work page arXiv 1967
[12]

Universality, characteristic kernels and rkhs embedding of measures

B. K. Sriperumbudur, K. Fukumizu, and G. R. Lanckriet, “Universality, characteristic kernels and rkhs embedding of measures.”Journal of Machine Learning Research, vol. 12, no. 7, 2011

work page 2011
[13]

Hilbert space embeddings and metrics on probability measures,

B. K. Sriperumbudur, A. Gretton, K. Fukumizu, B. Sch ¨olkopf, and G. R. Lanckriet, “Hilbert space embeddings and metrics on probability measures,”The Journal of Machine Learning Research, vol. 11, pp. 1517–1561, 2010

work page 2010
[14]

Robust estimation for independent non- homogeneous observations using density power divergence with applications to linear regression,

A. Ghosh and A. Basu, “Robust estimation for independent non- homogeneous observations using density power divergence with applications to linear regression,”Electronic Journal of Statistics, vol. 7, no. none, pp. 2420–2456, 2013. [Online]. Available: https://doi.org/10.1214/13-EJS847

work page doi:10.1214/13-ejs847 2013
[15]

Robust parametric classification and variable selection by a minimum distance criterion,

E. C. Chi and D. W. Scott, “Robust parametric classification and variable selection by a minimum distance criterion,”Journal of Computational and Graphical Statistics, vol. 23, no. 1, pp. 111–128, 2014

work page 2014
[16]

Universal robust regression via maximum mean discrepancy,

P. Alquier and M. Gerber, “Universal robust regression via maximum mean discrepancy,”Biometrika, vol. 111, no. 1, pp. 71–92, 05 2023. [Online]. Available: https://doi.org/10.1093/biomet/asad031

work page doi:10.1093/biomet/asad031 2023
[17]

Parametric statistical modeling by minimum integrated square error,

D. W. Scott, “Parametric statistical modeling by minimum integrated square error,”Technometrics, vol. 43, no. 3, pp. 274–285, 2001. [Online]. Available: https://doi.org/10.1198/004017001316975880

work page doi:10.1198/004017001316975880 2001
[18]

The l2e method,

——, “The l2e method,”WIREs Computational Statistics, vol. 1, no. 1, pp. 45–51, 2009. [Online]. Available: https://wires.onlinelibrary.wiley. com/doi/abs/10.1002/wics.4

work page doi:10.1002/wics.4 2009
[19]

Regularization paths for generalized linear models via coordinate descent,

J. H. Friedman, T. Hastie, and R. Tibshirani, “Regularization paths for generalized linear models via coordinate descent,”Journal of Statistical Software, vol. 33, no. 1, p. 1–22, 2010. [Online]. Available: https://www.jstatsoft.org/index.php/jss/article/view/v033i01

work page 2010
[20]

Lee and J

J. BARZILAI and J. M. BORWEIN, “Two-point step size gradient methods,”IMA Journal of Numerical Analysis, vol. 8, no. 1, pp. 141–148, 01 1988. [Online]. Available: https://doi.org/10.1093/imanum/ 8.1.141

work page doi:10.1093/imanum/ 1988
[21]

On the convergence of block coordinate descent type methods,

A. Beck and L. Tetruashvili, “On the convergence of block coordinate descent type methods,”SIAM Journal on Optimization, vol. 23, no. 4, pp. 2037–2060, 2013

work page 2037
[22]

Friedman, T

J. Friedman, T. Hastie, R. Tibshirani, B. Narasimhan, K. Tay, N. Simon, and J. Yang,glmnet: Lasso and Elastic-Net Regularized Generalized Linear Models, 2025, r package version 4.1-10. [Online]. Available: https://glmnet.stanford.edu

work page 2025
[23]

Li and Y

S. Li and Y . Qin,MTE: Maximum Tangent Likelihood Estimation for Robust Linear Regression and V ariable Selection, 2025, r package version 1.2.1. [Online]. Available: https://CRAN.R-project.org/package= MTE

work page 2025
[24]

robustHD: An R package for robust regression with high- dimensional data,

A. Alfons, “robustHD: An R package for robust regression with high- dimensional data,”Journal of Open Source Software, vol. 6, no. 67, p. 3786, 2021

work page 2021

[1] [1]

Fair: Facilitating artifi- cial intelligence resilience in manufacturing industrial internet,

Y . Zeng, I. Lourentzou, X. Deng, and R. Jin, “Fair: Facilitating artifi- cial intelligence resilience in manufacturing industrial internet,”arXiv preprint arXiv:2503.01086, 2025

work page arXiv 2025

[2] [2]

Robust estimation of a location parameter,

P. J. Huber, “Robust estimation of a location parameter,”The Annals of Mathematical Statistics, vol. 35, no. 1, pp. 73–101, 1964

work page 1964

[3] [3]

Asymptotic normality of r-estimates in the linear model,

S. Heiler and R. Willers, “Asymptotic normality of r-estimates in the linear model,”Statistics, vol. 19, no. 2, pp. 173–184, 1988. [Online]. Available: https://doi.org/10.1080/02331888808802084

work page doi:10.1080/02331888808802084 1988

[4] [4]

Robust regression by means of s- estimators,

P. Rousseeuw and V . Yohai, “Robust regression by means of s- estimators,” inRobust and Nonlinear Time Series Analysis, J. Franke, W. H¨ardle, and D. Martin, Eds. New York, NY: Springer US, 1984, pp. 256–272

work page 1984

[5] [5]

Robust linear model selection based on least angle regression,

J. A. Khan, S. V . Aelst, and R. H. Zamar, “Robust linear model selection based on least angle regression,”Journal of the American Statistical Association, vol. 102, no. 480, pp. 1289–1299, 2007

work page 2007

[6] [6]

Robust groupwise least angle regression,

A. Alfons, C. Croux, and S. Gelper, “Robust groupwise least angle regression,”Computational Statistics & Data Analysis, vol. 93, pp. 421–435, 2016. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S0167947315000468

work page 2016

[7] [7]

Robust lasso regression using tukey’s biweight criterion,

L. Chang, S. Roberts, and A. Welsh, “Robust lasso regression using tukey’s biweight criterion,”Technometrics, vol. 60, no. 1, pp. 36–47, 2018

work page 2018

[8] [8]

Statistical consistency and asymptotic normality for high-dimensional robustM-estimators,

P.-L. Loh, “Statistical consistency and asymptotic normality for high-dimensional robustM-estimators,”The Annals of Statistics, vol. 45, no. 2, pp. 866–896, 2017. [Online]. Available: https: //doi.org/10.1214/16-AOS1471

work page doi:10.1214/16-aos1471 2017

[9] [9]

Robust and efficient estimation by minimising a density power divergence,

A. Basu, I. R. Harris, N. L. Hjort, and M. Jones, “Robust and efficient estimation by minimising a density power divergence,”Biometrika, vol. 85, no. 3, pp. 549–559, 1998

work page 1998

[10] [10]

Robust parameter estimation with a small bias against heavy contamination,

H. Fujisawa and S. Eguchi, “Robust parameter estimation with a small bias against heavy contamination,”Journal of Multivariate Analysis, vol. 99, no. 9, pp. 2053–2081, 2008

work page 2053

[11] [11]

The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming,

L. Bregman, “The relaxation method of finding the common point of convex sets and its application to the solution of problems in convex programming,”USSR Computational Mathematics and Mathematical Physics, vol. 7, no. 3, pp. 200–217, 1967. [Online]. Available: https://www.sciencedirect.com/science/article/pii/0041555367900407

work page arXiv 1967

[12] [12]

Universality, characteristic kernels and rkhs embedding of measures

B. K. Sriperumbudur, K. Fukumizu, and G. R. Lanckriet, “Universality, characteristic kernels and rkhs embedding of measures.”Journal of Machine Learning Research, vol. 12, no. 7, 2011

work page 2011

[13] [13]

Hilbert space embeddings and metrics on probability measures,

B. K. Sriperumbudur, A. Gretton, K. Fukumizu, B. Sch ¨olkopf, and G. R. Lanckriet, “Hilbert space embeddings and metrics on probability measures,”The Journal of Machine Learning Research, vol. 11, pp. 1517–1561, 2010

work page 2010

[14] [14]

Robust estimation for independent non- homogeneous observations using density power divergence with applications to linear regression,

A. Ghosh and A. Basu, “Robust estimation for independent non- homogeneous observations using density power divergence with applications to linear regression,”Electronic Journal of Statistics, vol. 7, no. none, pp. 2420–2456, 2013. [Online]. Available: https://doi.org/10.1214/13-EJS847

work page doi:10.1214/13-ejs847 2013

[15] [15]

Robust parametric classification and variable selection by a minimum distance criterion,

E. C. Chi and D. W. Scott, “Robust parametric classification and variable selection by a minimum distance criterion,”Journal of Computational and Graphical Statistics, vol. 23, no. 1, pp. 111–128, 2014

work page 2014

[16] [16]

Universal robust regression via maximum mean discrepancy,

P. Alquier and M. Gerber, “Universal robust regression via maximum mean discrepancy,”Biometrika, vol. 111, no. 1, pp. 71–92, 05 2023. [Online]. Available: https://doi.org/10.1093/biomet/asad031

work page doi:10.1093/biomet/asad031 2023

[17] [17]

Parametric statistical modeling by minimum integrated square error,

D. W. Scott, “Parametric statistical modeling by minimum integrated square error,”Technometrics, vol. 43, no. 3, pp. 274–285, 2001. [Online]. Available: https://doi.org/10.1198/004017001316975880

work page doi:10.1198/004017001316975880 2001

[18] [18]

The l2e method,

——, “The l2e method,”WIREs Computational Statistics, vol. 1, no. 1, pp. 45–51, 2009. [Online]. Available: https://wires.onlinelibrary.wiley. com/doi/abs/10.1002/wics.4

work page doi:10.1002/wics.4 2009

[19] [19]

Regularization paths for generalized linear models via coordinate descent,

J. H. Friedman, T. Hastie, and R. Tibshirani, “Regularization paths for generalized linear models via coordinate descent,”Journal of Statistical Software, vol. 33, no. 1, p. 1–22, 2010. [Online]. Available: https://www.jstatsoft.org/index.php/jss/article/view/v033i01

work page 2010

[20] [20]

Lee and J

J. BARZILAI and J. M. BORWEIN, “Two-point step size gradient methods,”IMA Journal of Numerical Analysis, vol. 8, no. 1, pp. 141–148, 01 1988. [Online]. Available: https://doi.org/10.1093/imanum/ 8.1.141

work page doi:10.1093/imanum/ 1988

[21] [21]

On the convergence of block coordinate descent type methods,

A. Beck and L. Tetruashvili, “On the convergence of block coordinate descent type methods,”SIAM Journal on Optimization, vol. 23, no. 4, pp. 2037–2060, 2013

work page 2037

[22] [22]

Friedman, T

J. Friedman, T. Hastie, R. Tibshirani, B. Narasimhan, K. Tay, N. Simon, and J. Yang,glmnet: Lasso and Elastic-Net Regularized Generalized Linear Models, 2025, r package version 4.1-10. [Online]. Available: https://glmnet.stanford.edu

work page 2025

[23] [23]

Li and Y

S. Li and Y . Qin,MTE: Maximum Tangent Likelihood Estimation for Robust Linear Regression and V ariable Selection, 2025, r package version 1.2.1. [Online]. Available: https://CRAN.R-project.org/package= MTE

work page 2025

[24] [24]

robustHD: An R package for robust regression with high- dimensional data,

A. Alfons, “robustHD: An R package for robust regression with high- dimensional data,”Journal of Open Source Software, vol. 6, no. 67, p. 3786, 2021

work page 2021