Pith. sign in

REVIEW 4 major objections 6 minor 35 references

Iterative Reweighted Framework Based Algorithms for Sparse Linear Regression with Generalized Elastic Net Penalty

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that the generalized elastic net with an ℓr-norm loss and an ℓq-norm penalty (0<q<1) is solvable by two iterative reweighted algorithms whose accumulation points are generalized first-order stationary points.

desk verdict Solid incremental methods paper whose main theorems are fine but whose stated optimality corollary and experiment-theory connection both need fixing. read the letter →

arxiv 2411.14875 v2 pith:RT3OOYMG submitted 2024-11-22 stat.ML cs.LGmath.STstat.TH

Yanyun Ding , Zhenghua Yao , Peili Li , Yunhai Xiao This is my paper · ORCID
classification stat.MLcs.LGmath.STstat.TH MSC 62J0790C2690C25
keywords sparselinearregressiongeneralizedelasticnetpenaltyiterativereweightedframeworkalternatingdirectionmethodofmultipliersproximalmajorization-minimizationsemismoothNewtonℓq-normℓr-normloss
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

The paper is trying to establish that the generalized elastic-net model $\min_{\beta\in\mathbb{R}^p}\{ \|X\beta-y\|_r + \lambda_2\|\beta\|_2^2 + \lambda_1\|\beta\|_q^q \}$, with $r\ge 1$ and $0

What carries the argument

The load-bearing object is the $\varepsilon$-approximation $h_{u_\varepsilon}(\beta_i)=\min_{0\le s\le u_\varepsilon} q\left(|\beta_i|s - \frac{q-1}{q}s^{q/(q-1)}\right)$ with $u_\varepsilon=(\varepsilon/(p\lambda_1))^{(q-1)/q}$; it is Lipschitz, has a computable Clarke subdifferential, and converts the $\ell_q^q$ term into an iteratively reweighted $\ell_1$ term with weights $w_i^k=\min\{u_\varepsilon,|\beta_i^k|^{q-1}\}$. Around it sit the proximal mappings of the $\ell_r$ norms for $r=1,2,\infty$, which are strongly semismooth and give closed-form updates in ADMM and a well-defined generalized Hessian in the semismooth Newton step. The threshold (2.8), involving $\|X_i\|_r$, $\lambda_1$, and $q$, is what makes a stationary point of the approximation a stationary point of the original model.

What would settle it

Run the one-predictor case ($p=1$, $q=1/2$, fixed $\lambda_1$) with random noise, compute the threshold (2.8) from $\|X_1\|_r$, and run the reweighted algorithm with $\varepsilon$ set above that threshold. If the accumulation point violates the lower bound $|\bar{\beta}_1|>(q\lambda_1/\|X_1\|_r)^{1/(1-q)}$, or fails the stationarity condition (2.1), the claim that the implemented algorithm is covered by Theorem 4 would be falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that the nonconvex, non-Lipschitz problem (1.3) can be approximated by a convex reweighted $\ell_1$ problem at each outer iteration, and that the approximation is faithful in a precise sense. Replacing $\|\beta\|_q^q$ with the Lipschitz $\varepsilon$-approximation $h_{u_\varepsilon}$ turns the objective into the convex problem (2.7); the iterative reweighted scheme then solves (3.2), whose objective is the $\ell_r$-loss plus a weighted $\ell_1$ term. Theorem 4 states that if $\varepsilon$ satisfies (2.8), any accumulation point $\bar{\beta}$ of the sequence is a generalized first-order stationary point of the original model in the sense of (2.1), and its nonzero entries satisfy $|\bar{\beta}_i|>(q\lambda_1/\|X_i\|_r)^{1/(1-q)}$. The same framework underlies both algorithms: ADMM solves the convex subproblem with closed-form proximal updates for $\ell_r$ norms, while PMM-SSN solves the subproblem through its dual with a semismooth Newton method. The numerical section claims that both are competitive with an existing generalized-elastic-net solver and that PMM-SSN is more efficient than ADMM.

Load-bearing premise

The convergence and equivalence theorems hold only when the smoothing parameter $\varepsilon$ is below the threshold in (2.8), which depends on the column norms $\|X_i\|_r$ and on $\lambda_1$ and $q$; the numerical experiments never report how $\varepsilon$ is set, so the implemented algorithms may be running outside the regime where their accumulation points are provably stationary for the original model.

Editorial extensions

If this is right

  • Every solution returned by either algorithm has no nonzero coefficient smaller than $(q\lambda_1/\|X_i\|_r)^{1/(1-q)}$, giving a direct threshold for screening variables out of the support.
  • With $r=1$, $r=2$, and $r=\infty$ the same algorithmic pair covers least-absolute-deviation, square-root, and uniform-noise losses without changing the stationarity guarantees.
  • The outer sequence decreases the $\varepsilon$-approximation objective monotonically, so early termination based on relative change preserves the stationarity statement.
  • On the tested instances PMM-SSN reaches the stopping tolerance in fewer iterations than ADMM, and both outperform the IAGENR-Lq baseline in mean squared error and relative error.

Reading between the lines

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

  • The paper gives no default value for $\varepsilon$; a natural extension is a data-dependent rule that sets $\varepsilon$ as a fixed small fraction of the right-hand side of (2.8), making the guarantee hold by construction for every problem instance.
  • The lower bound (2.4) can be used as a post-hoc diagnostic: any reported coefficient below the bound is a warning that the solver stopped outside the stationary regime or that the model parameters are mis-specified.
  • If the stationarity argument only needs the loss's proximal map to be strongly semismooth, the same reweighted outer loop should extend to other losses such as quantile or Huber, as long as their proximal operators are available.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. This paper studies the sparse linear regression model F(β)=||Xβ−y||_r+λ2||β||_2^2+λ1||β||_q^q with r≥1 and 0<q<1. It defines a generalized first-order stationarity condition, proves that local minimizers satisfy it, and derives a lower bound on the magnitude of nonzero entries of stationary points. It then replaces the ℓ_q penalty by Lu's locally Lipschitz ε-approximation, obtains a convex reweighted subproblem, and proposes ADMM and PMM-SSN algorithms for it. The main theoretical claim is that accumulation points of the exact reweighted iteration (3.2) are generalized stationary points of (1.3) when ε satisfies (2.8), and the numerical sections report experiments on simulated and real data.

Significance. If the convergence claims hold, the paper provides a useful extension of the iterative reweighted framework to ℓ_r-loss elastic-net problems with nonconvex ℓ_q penalties, and the lower-bound result is a concrete identifiability statement. The PMM-SSN dual approach, which exploits strongly semismooth proximal mappings, is a sensible and potentially efficient design, and the numerical study compares with a recent alternative. However, the significance is conditional: the implemented algorithms are not connected to the convergence theorem by the reported parameters, and one corollary asserts an unproved direction. No machine-checked proofs or code are provided, so verification rests on the written proofs and tables.

major comments (4)
  1. [§2.3, Corollary 2.1] The statement that an optimal solution of (1.3) is also an optimal solution of (2.7) is not a consequence of Theorems 1 and 3. Theorem 3 only shows that stationary points of (2.7) are stationary for (1.3), and Theorem 1 only shows that local minimizers of (1.3) are stationary for (1.3); neither gives the asserted direction. Please correct the statement or provide a separate proof.
  2. [§3, Theorem 4; §4.1] The convergence guarantee of Theorem 4 requires ε to satisfy the column-dependent threshold (2.8), but Section 4 never reports the value of ε used in any experiment and never checks (2.8) for the reported data and parameters. Without this information, the numerical solutions are only proved to be stationary points of the ε-approximation (2.7) under unverified conditions, not of the original model (1.3). The experiments should state ε and verify the condition, or the claims should be weakened accordingly.
  3. [§3, Theorem 4 and Algorithms 1–2] Theorem 4 assumes that each outer iterate β^{k+1} is an exact minimizer of the reweighted problem (3.1), while both algorithms stop the inner iterations at η2≤1e−4 (or η1≤1e−4 in §4.4). The theory therefore does not cover the implemented algorithms. Please add an inexact-version convergence theorem with explicit tolerance conditions, or justify that the chosen tolerances place the iterates in the theorem's regime.
  4. [§3.1] The convergence of the ADMM for (3.1) is attributed to references [33,34], but neither reference establishes the needed two-block ADMM convergence for this problem: [33] is a counterexample paper on multi-block ADMM, and [34] concerns Hankel matrix rank minimization. Please give a precise convergence statement and a correct citation, or prove the result.
minor comments (6)
  1. [§2.1] In the first paragraph, the complement \bar T is taken in {1,...,n}, but β∈R^p; it should be {1,...,p}.
  2. [§4.2.1] The fixed-data experiment sets q=1, which is the ordinary elastic-net penalty rather than the 0<q<1 generalized penalty studied in the paper; please explain why this case is included or move it to a baseline.
  3. [Table 1] In the UN, λ1=0.05 row, ADMM reports MSE=3.432e−01 and RE=1.129e−01, substantially worse than the neighboring rows; this apparent failure should be discussed rather than summarized as consistent performance.
  4. [Algorithm 2] The displayed update for β^{k+1} is typeset ambiguously and should include the factor (2λ2)^{-1} explicitly, as in Prox_{(2λ2)^{-1}g_k(·)}(−(2λ2)^{-1}X^T \bar u^{k+1}).
  5. [Throughout] There are numerous typographical errors (e.g., "miminizes", "resuts", "Thorem", "PMM-SSN is efficient than ADMM"); a careful proofread is needed.
  6. [§4.4] The LIBSVM URL given for the regression datasets points to a binary classification page; please supply the correct URLs or dataset descriptions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the convergence results are derived from first-order optimality conditions and the epsilon-approximation, not from fitting or from the paper's own conclusions.

full rationale

The paper's central theoretical claim is that accumulation points of the exact iterative reweighted scheme (3.2) are generalized first-order stationary points of the original model (1.3), provided the epsilon threshold (2.8) holds. This is proven in Theorem 4 by showing that each accumulation point is stationary for the convex epsilon-approximation (2.7), via the descent inequality (3.4) and the first-order condition (3.6), and then transferring the stationarity to the original model through Theorem 3. Theorem 3's proof uses condition (2.8) only to force the nonzero entries of a stationary point of the approximation out of the linear region of h_u^epsilon, so that partial h_u^epsilon coincides with the true lq derivative; this is an analytic argument, not a restatement of the conclusion. Theorem 2 derives the lower bound algebraically from the stationarity definition (2.1), and Theorem 1 is a standard local-minimizer-to-Clarke-stationarity implication. No equation in the paper defines a quantity in terms of the quantity it purports to predict, and the numerical comparisons, while based on the authors' own tuning choices, are not presented as predictions forced by the theory. The self-citations [18] and [21] supply standard semismoothness and proximal-mapping facts and are not load-bearing for the convergence theorems. The gap between the implemented algorithms and the exact iteration, and the unreported epsilon values in Section 4, are correctness or reproducibility concerns, not circularity. Accordingly, no circular step is identified; the derivation is self-contained relative to its stated assumptions.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The theoretical derivation adds no new fitted constants, but the empirical claims depend on hand-tuned model and algorithm parameters. The key unstated premise is the threshold condition on epsilon, which is required by the convergence theorems but never instantiated in the experiments.

free parameters (4)
  • q (penalty exponent) = 0.5 in main experiments
    Chosen after comparing q=0.2,...,0.9 in Section 4.2.2; the paper sets q=0.5 based on those same experiments, a post hoc selection.
  • lambda1, lambda2 (tuning parameters) = varies by noise, e.g., lambda1=0.2, lambda2=1e-4
    Selected per experiment through extensive tests in Section 4.1; these hand-tuned values affect the reported performance comparisons.
  • Algorithm parameters (sigma, tau, mu, rho, delta, nu) = fixed values listed in Sections 4.2 and 4.4
    Hand-chosen constants for ADMM and PMM-SSN; the claimed efficiency advantage depends on these choices.
  • epsilon (approximation parameter) = not specified in experiments
    The convergence guarantee requires condition (2.8), but the paper does not state how epsilon is set in the numerical tests.
assumptions (5)
  • standard math The ℓr-norm proximal mapping is strongly semismooth and the generalized Hessian formula for r=1,2,∞ holds.
    Imported from references [18,28] and used in the SSN part of PMM-SSN in Section 3.2.
  • standard math The epsilon approximation h_u_epsilon from Lu has the subdifferential (2.6) and supports the equivalence between (2.7) and (1.3).
    Taken from Lu [27] and used throughout Theorems 3 and 4.
  • standard math The dual function Θ(u) is convex and continuously differentiable with gradient given by the Moreau envelope identity.
    Standard convex analysis result used in Section 3.2 to derive the SSN method.
  • domain assumption The data satisfy y=Xβ+ε with β sparse, and the noise type is modeled by choosing r in {1,2,∞}.
    The model and all experiments are built on this high-dimensional sparse linear regression assumption from Section 1.
  • domain assumption The threshold condition (2.8) is satisfied by the chosen epsilon.
    Theorems 3 and 4 require this condition, but the numerical section does not specify epsilon values.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Iterative Reweighted Framework Based Algorithms for Sparse Linear Regression with Generalized Elastic Net Penalty." pith.science (2026). https://pith.science/paper/RT3OOYMG

@misc{pith2026241114875,
  author       = {Pith},
  title        = {Pith review of: Iterative Reweighted Framework Based Algorithms for Sparse Linear Regression with Generalized Elastic Net Penalty},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RT3OOYMG}},
  note         = {Machine review of arXiv:2411.14875}
}
abstract

The elastic net penalty is frequently employed in high-dimensional statistics for parameter regression and variable selection. It is particularly beneficial compared to lasso when the number of predictors greatly surpasses the number of observations. However, empirical evidence has shown that the $\ell_q$-norm penalty (where $0 < q < 1$) often provides better regression compared to the $\ell_1$-norm penalty, demonstrating enhanced robustness in various scenarios. In this paper, we explore a generalized elastic net model that employs a $\ell_r$-norm (where $r \geq 1$) in loss function to accommodate various types of noise, and employs a $\ell_q$-norm (where $0 < q < 1$) to replace the $\ell_1$-norm in elastic net penalty. Theoretically, we establish the computable lower bounds for the nonzero entries of the generalized first-order stationary points of the proposed generalized elastic net model. For implementation, we develop two efficient algorithms based on the locally Lipschitz continuous $\epsilon$-approximation to $\ell_q$-norm. The first algorithm employs an alternating direction method of multipliers (ADMM), while the second utilizes a proximal majorization-minimization method (PMM), where the subproblems are addressed using the semismooth Newton method (SNN). We also perform extensive numerical experiments with both simulated and real data, showing that both algorithms demonstrate superior performance. Notably, the PMM-SSN is efficient than ADMM, even though the latter provides a simpler implementation.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [33]

    Ma thematical Programming 155(1), 57–79 (2016)

    Chen, C., He, B., Ye, Y., Yuan, X.: The direct extension of admm fo r multi- block convex minimization problems is not necessarily convergent. Ma thematical Programming 155(1), 57–79 (2016)

    work page 2016

  2. [34]

    SIAM Journal o n Matrix Analysis and Applications 34(3), 946–977 (2013)

    Fazel, M., Pong, T.-K., Sun, D., Tseng, P.: Hankel matrix rank minim ization with applications to system identification and realization. SIAM Journal o n Matrix Analysis and Applications 34(3), 946–977 (2013)

    work page 2013

  3. [1]

    SIAM review 43(1), 129–159 (2001) 24

    Chen, S.S., Donoho, David, L., Saunders, Michael, A.: Atomic decom position by basis pursuit. SIAM review 43(1), 129–159 (2001) 24

    work page 2001

  4. [2]

    Jo urnal of the Royal Statistical Society Series B 58(1), 267–288 (1996)

    Tibshirani, R.: Regression shrinkage and selection via the lasso. Jo urnal of the Royal Statistical Society Series B 58(1), 267–288 (1996)

    work page 1996

  5. [3]

    Annals of Statistics (2), 32 (2004)

    Efron, B., Hastie, T., Johnstone, I., Tibshirani, R.: Least angle re gression. Annals of Statistics (2), 32 (2004)

    work page 2004

  6. [4]

    Annals of Statistics 37(1), 246–270 (2009)

    Meinshausen, N., Yu, B.: Lasso-type recovery of sparse repre sentations for high- dimensional data. Annals of Statistics 37(1), 246–270 (2009)

    work page 2009

  7. [5]

    Journal of the Royal Statistical Society Series B 67(2), 301–320 (2005)

    Zou, H., Hastie, T.: Regularization and variable selection via the elas tic net. Journal of the Royal Statistical Society Series B 67(2), 301–320 (2005)

    work page 2005

  8. [6]

    IEEE Signal Processing Letters 14(10), 707–710 (2007)

    Rick, C.: Exact reconstruction of sparse signals via nonconvex m inimization. IEEE Signal Processing Letters 14(10), 707–710 (2007)

    work page 2007

Show all 35 references
  1. [7]

    IEEE transactions on signal processing 59(10), 5086–5090 (2011)

    Sun, Q.: Sparse approximation property and stable recovery of sparse signals from noisy measurements. IEEE transactions on signal processing 59(10), 5086–5090 (2011)

    2011

  2. [8]

    Advances in Pure Mathematics 7(1), 66–74 (2017)

    Li, S., Ye, W.: A generalized elastic net regularization with smoothed l0 penalty. Advances in Pure Mathematics 7(1), 66–74 (2017)

    2017

  3. [9]

    Annals of Statistics 39(1), 82–130 (2011)

    Belloni, A., Chernozhukov, V.: ℓ1-penalized quantile regression in high- dimensional sparse models. Annals of Statistics 39(1), 82–130 (2011)

    2011

  4. [10]

    Journal of the Royal Statistical So ciety Series B 73(3), 325–349 (2011)

    Bradic, J., Fan, J., Wang, W.: Penalized composite quasi-likelihood f or ultrahigh dimensional variable selection. Journal of the Royal Statistical So ciety Series B 73(3), 325–349 (2011)

    2011

  5. [11]

    A nnals of Statistics 42(1), 324–351 (2012)

    Fan, J., Fan, Y., Barut, E.: Adaptive robust variable selection. A nnals of Statistics 42(1), 324–351 (2012)

    2012

  6. [12]

    Journal of Multivariate Analysis 120, 135–151 (2013)

    Wang, L.: The ℓ1 penalized lad estimator for high dimensional linear regression. Journal of Multivariate Analysis 120, 135–151 (2013)

    2013

  7. [13]

    Journal of the American Statistical Association 107(497), 214– 222 (2012)

    Wang, L., Wu, Y., Li, R.: Quantile regression for analyzing heterog eneity in ultra- high dimension. Journal of the American Statistical Association 107(497), 214– 222 (2012)

    2012

  8. [14]

    Biometrika 98(4), 791–806 (2011)

    Belloni, A., Chernozhukov, V., Wang, L.: Square-root lasso: pivo tal recovery of sparse signals via conic programming. Biometrika 98(4), 791–806 (2011)

    2011

  9. [15]

    Annals of Statistics 46(6B), 3603–3642 (2016)

    Bellec, P.C., Lecu´ e, G., Tsybakov, A.B.: Slope meets lasso: improv ed oracle bounds and optimality. Annals of Statistics 46(6B), 3603–3642 (2016)

    2016

  10. [16]

    Electronic Journal of Statistics 12(1), 741–766 (2018) 25

    Derumigny, A.: Improved bounds for square-root lasso and sq uare-root slope. Electronic Journal of Statistics 12(1), 741–766 (2018) 25

    2018

  11. [17]

    Journal of Machine Learning Research 21(226), 1–38 (2020)

    Tang, P., Wang, C., Sun, D., Toh, K.-C.: A sparse semismooth newt on based proximal majorization-minimization algorithm for nonconvex square -root-loss regression problems. Journal of Machine Learning Research 21(226), 1–38 (2020)

    2020

  12. [18]

    Optimization Methods a nd Software 38(2), 262–288 (2023)

    Ding, Y., Zhang, H., Li, P., Xiao, Y.: An efficient semismooth newton m ethod for adaptive sparse signal recovery problems. Optimization Methods a nd Software 38(2), 262–288 (2023)

    2023

  13. [19]

    Journal of Scientific Computing 75(2), 713–732 (2018)

    Wen, Y.-W., Ching, W.-K., Ng, M.: A semi-smooth newton method for inverse problem with uniform noise. Journal of Scientific Computing 75(2), 713–732 (2018)

    2018

  14. [20]

    Communications in Mathematica l Sciences 18, 227–258 (2020)

    Xue, Y., Feng, Y., Wu, C.: An efficient and globally convergent algor ithm for ℓp,q-ℓr model in group sparse optimizationn. Communications in Mathematica l Sciences 18, 227–258 (2020)

    2020

  15. [21]

    Journal of Nonlinear & Variational Analysis 8(3), 433–449 (2024)

    Xiao, Y., Shen, J., Ding, Y., Shi, M., Li, P.: A fast and effective algorit hm for sparse linear regression with lp-norm data fidelity and elastic net reg ularization. Journal of Nonlinear & Variational Analysis 8(3), 433–449 (2024)

    2024

  16. [22]

    Mathematical Programming 143(1-2), 371–383 (2011)

    Chen, X., Ge, D., Wang, Z., Ye, Y.: Complexity of unconstrained ℓ2-ℓp minimiza- tion. Mathematical Programming 143(1-2), 371–383 (2011)

    2011

  17. [23]

    Mathematical Programming 129(2), 285–299 (2011)

    Ge, D., Jian, X., Ye, Y.: A note on the complexity of ℓp minimization. Mathematical Programming 129(2), 285–299 (2011)

    2011

  18. [24]

    Computational Optimization and Applications 59(1-2), 47–61 (2014)

    Chen, X., Zhou, W.: Convergence of the reweighted ℓ1 minimization algorithm for ℓ2-ℓp minimization. Computational Optimization and Applications 59(1-2), 47–61 (2014)

    2014

  19. [25]

    Communications on Pure and Applied Mathematics: A Journal Issued by the Courant Institute o f Mathematical Sciences 63(1), 1–38 (2010)

    Daubechies, I., DeVore, R., Fornasier, M., G¨ unt¨ urk, C.S.: Iteratively reweighted least squares minimization for sparse recovery. Communications on Pure and Applied Mathematics: A Journal Issued by the Courant Institute o f Mathematical Sciences 63(1), 1–38 (2010)

    2010

  20. [26]

    SIAM Journal on Optimization 21(1), 82–101 (2011)

    Lai, M.J., Wang, J.: An unconstrained ℓq minimization with 0 ≤ q ≤ 1 for sparse solution of underdetermined linear systems. SIAM Journal on Optimization 21(1), 82–101 (2011)

    2011

  21. [27]

    Mathematical Programming 147(1–2), 277–307 (2014)

    Lu, Zhaosong: Iterative reweighted minimization methods for ℓp regularized unconstrained nonlinear programming. Mathematical Programming 147(1–2), 277–307 (2014)

    2014

  22. [28]

    Mathematical Programming Computation 14(2), 223–270 (2022) 26

    Lin, M., Sun, D., Toh, K.-C.: An augmented lagrangian method with c onstraint generation for shape-constrained convex regression problems. Mathematical Programming Computation 14(2), 223–270 (2022) 26

    2022

  23. [29]

    Hiriart-Urruty, J.B., Lemar´ echal, C.: Convex Analysis and Minimiz ation Algo- rithms I: Fundamentals, Berlin Heidelberg

  24. [30]

    SIAM Journal on Optimization 7(2), 367–385 (1997)

    Lemar´ echal, C., Sagastiz´ abal, C.: Practical aspects of the moreau–yosida regular- ization: Theoretical preliminaries. SIAM Journal on Optimization 7(2), 367–385 (1997)

    1997

  25. [31]

    Rockafellar, R.T.: Convex Analysis, Princeton

  26. [32]

    Computational Optimization and Applications 70(1), 201–219 (2018)

    Xiu, X., Kong, L., Li, Y., Qi, H.: Iterative reweighted methods for ℓ1-ℓp minimization. Computational Optimization and Applications 70(1), 201–219 (2018)

    2018

  27. [35]

    Advances in neural information processing systems (23) (2010) 27

    Huang, L., Jia, J., Yu, B., Chun, B.-G., Maniatis, P., Naik, M.: Predict ing exe- cution time of computer programs using sparse polynomial regress ion. Advances in neural information processing systems (23) (2010) 27

    2010

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.