Globally-convergent Iteratively Reweighted Least Squares for Robust Regression Problems

被引:0
|
作者
Mukhoty, Bhaskar [1 ]
Gopakumar, Govind [1 ]
Jain, Prateek [2 ]
Kar, Purushottam [1 ]
机构
[1] IIT Kanpur, Kanpur, Uttar Pradesh, India
[2] Microsoft Res India, Bengaluru, India
关键词
MINIMIZATION;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We provide the first global model recovery results for the IRLS (iteratively reweighted least squares) heuristic for robust regression problems. IRLS is known to offer excellent performance, despite bad initializations and data corruption, for several parameter estimation problems. Existing analyses of IRLS frequently require careful initialization, thus offering only local convergence guarantees. We remedy this by proposing augmentations to the basic IRLS routine that not only offer guaranteed global recovery, but in practice also outperform state-of-the-art algorithms for robust regression. Our routines are more immune to hyperparameter misspecification in basic regression tasks, as well as applied tasks such as linear-armed bandit problems. Our theoretical analyses rely on a novel extension of the notions of strong convexity and smoothness to weighted strong convexity and smoothness, and establishing that sub-Gaussian designs offer bounded weighted condition numbers. These notions may be useful in analyzing other algorithms as well.
引用
收藏
页码:313 / 322
页数:10
相关论文
共 50 条
  • [1] ROBUST REGRESSION COMPUTATION USING ITERATIVELY REWEIGHTED LEAST-SQUARES
    OLEARY, DP
    [J]. SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 1990, 11 (03) : 466 - 480
  • [2] Robust spectrotemporal decomposition by iteratively reweighted least squares
    Ba, Demba
    Babadi, Behtash
    Purdon, Patrick L.
    Brown, Emery N.
    [J]. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2014, 111 (50) : E5336 - E5345
  • [3] Robust regularized extreme learning machine for regression using iteratively reweighted least squares
    Chen, Kai
    Lv, Qi
    Lu, Yao
    Dou, Yong
    [J]. NEUROCOMPUTING, 2017, 230 : 345 - 358
  • [4] A NOTE ON COMPUTING ROBUST REGRESSION ESTIMATES VIA ITERATIVELY REWEIGHTED LEAST-SQUARES
    STREET, JO
    CARROLL, RJ
    RUPPERT, D
    [J]. AMERICAN STATISTICIAN, 1988, 42 (02): : 152 - 154
  • [6] Robust registration of point sets using iteratively reweighted least squares
    Per Bergström
    Ove Edlund
    [J]. Computational Optimization and Applications, 2014, 58 : 543 - 561
  • [7] Robust registration of point sets using iteratively reweighted least squares
    Bergstrom, Per
    Edlund, Ove
    [J]. COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2014, 58 (03) : 543 - 561
  • [8] Kernel-based regression via a novel robust loss function and iteratively reweighted least squares
    Dong, Hongwei
    Yang, Liming
    [J]. KNOWLEDGE AND INFORMATION SYSTEMS, 2021, 63 (05) : 1149 - 1172
  • [9] Kernel-based regression via a novel robust loss function and iteratively reweighted least squares
    Hongwei Dong
    Liming Yang
    [J]. Knowledge and Information Systems, 2021, 63 : 1149 - 1172
  • [10] Iteratively reweighted least squares based learning
    Warner, BA
    Misra, M
    [J]. IEEE WORLD CONGRESS ON COMPUTATIONAL INTELLIGENCE, 1998, : 1327 - 1331