Subsampled cubic regularization method for finite-sum minimization

被引：0

作者：

Goncalves, Max L. N. ^{[1
]}

机构：

[1] Univ Fed Goias, IME, Goiania, Brazil

来源：

OPTIMIZATION | 2025年 / 74卷 / 07期

关键词：

Cubic regularization method; subsampling strategy; iteration-complexity analysis; global convergence; finite-sum optimization problem; COMPLEXITY;

D O I：

10.1080/02331934.2024.2318258

中图分类号：

C93 [管理学]; O22 [运筹学];

学科分类号：

070105 ; 12 ; 1201 ; 1202 ; 120202 ;

摘要：

This paper proposes and analyses a subsampled Cubic Regularization Method (CRM) for solving finite-sum optimization problems. The new method uses random subsampling techniques to approximate the functions, gradients and Hessians in order to reduce the overall computational cost of the CRM. Under suitable hypotheses, first- and second-order iteration-complexity bounds and global convergence analyses are presented. We also discuss the local convergence properties of the method. Numerical experiments are presented to illustrate the performance of the proposed scheme.

引用

收藏

页码：1591 / 1614

页数：24

相关论文

共 26 条

[1] 2-POINT STEP SIZE GRADIENT METHODS [J].

BARZILAI, J ;

BORWEIN, JM .

IMA JOURNAL OF NUMERICAL ANALYSIS, 1988, 8 (01) :141-148

[2] Quadratic and Cubic Regularisation Methods with Inexact function and Random Derivatives for Finite-Sum Minimisation [J].

Bellavia, Stefania ;

Gurioli, Gianmarco ;

Morini, Benedetta ;

Toint, Philippe L. .

2021 21ST INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND ITS APPLICATIONS ICCSA 2021, 2021, :258-267

[3] Stochastic analysis of an adaptive cubic regularization method under inexact gradient evaluations and dynamic Hessian accuracy [J].

Bellavia, Stefania ;

Gurioli, Gianmarco .

OPTIMIZATION, 2022, 71 (01) :227-261

[4] Subsampled inexact Newton methods for minimizing large sums of convex functions [J].

Bellavia, Stefania ;

Krejic, Natasa ;

Jerinkic, Natasa Krklec .

IMA JOURNAL OF NUMERICAL ANALYSIS, 2020, 40 (04) :2309-2341

[5] Subsampled Nonmonotone Spectral Gradient Methods [J].

Bellavia, Stefania ;

Jerinkis, Natasa Krklec ;

Malaspina, Greta .

COMMUNICATIONS IN APPLIED AND INDUSTRIAL MATHEMATICS, 2020, 11 (01) :19-34

[6] Adaptive cubic regularization methods with dynamic inexact Hessian information and applications to finite-sum minimization [J].

Bellavia, Stefania ;

Gurioli, Gianmarco ;

Morini, Benedetta .

IMA JOURNAL OF NUMERICAL ANALYSIS, 2021, 41 (01) :764-799

[7] Inexact restoration with subsampled trust-region methods for finite-sum minimization [J].

Bellavia, Stefania ;

Krejic, Natasa ;

Morini, Benedetta .

COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2020, 76 (03) :701-736

[8] On the use of iterative methods in cubic regularization for unconstrained optimization [J].

Bianconcini, Tommaso ;

Liuzzi, Giampaolo ;

Morini, Benedetta ;

Sciandrone, Marco .

COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2015, 60 (01) :35-57

[9] Worst-case evaluation complexity for unconstrained nonlinear optimization using high-order regularized models [J].

Birgin, E. G. ;

Gardenghi, J. L. ;

Martinez, J. M. ;

Santos, S. A. ;

Toint, Ph. L. .

MATHEMATICAL PROGRAMMING, 2017, 163 (1-2) :359-368

[10] ON THE COMPLEXITY OF STEEPEST DESCENT, NEWTON'S AND REGULARIZED NEWTON'S METHODS FOR NONCONVEX UNCONSTRAINED OPTIMIZATION PROBLEMS [J].

Cartis, C. ;

Gould, N. I. M. ;

Toint, Ph. L. .

SIAM JOURNAL ON OPTIMIZATION, 2010, 20 (06) :2833-2852