Efficient Quadratic Penalization Through the Partial Minimization Technique

被引:21
作者
Aravkin, Aleksandr Y. [1 ]
Drusvyatskiy, Dmitriy [2 ]
van Leeuwen, Tristan [3 ]
机构
[1] Univ Washington, Dept Appl Math, Seattle, WA 98195 USA
[2] Univ Washington, Dept Math, Seattle, WA 98195 USA
[3] Univ Utrecht, Dept Math, NL-3584 CD Utrecht, Netherlands
关键词
Kalman smoothing; Nonconvex optimization; PDE constrained optimization; variable projection; ROBUST;
D O I
10.1109/TAC.2017.2754474
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Common computational problems, such as parameter estimation in dynamic models and partial differential equation (PDE)-constrained optimization, require data fitting over a set of auxiliary parameters subject to physical constraints over an underlying state. Naive quadratically penalized formulations, commonly used in practice, suffer from inherent ill-conditioning. We show that surprisingly the partial minimization technique regularizes the problem, making it well-conditioned. This viewpoint sheds newlight on variable projection techniques, as well as the penalty method for PDE-constrained optimization, and motivates robust extensions. In addition, we outline an inexact analysis, showing that the partial minimization subproblem can be solved very loosely in each iteration. We illustrate the theory and algorithms on boundary control, optimal transport, and parameter estimation for robust dynamic inference.
引用
收藏
页码:2131 / 2138
页数:8
相关论文
共 23 条
[1]   A User's Guide to Optimal Transport [J].
Ambrosio, Luigi ;
Gigli, Nicola .
MODELLING AND OPTIMISATION OF FLOWS ON NETWORKS, CETRARO, ITALY 2009, 2013, 2062 :1-155
[2]  
ANDERSON B. D., 2012, Optimal Filtering
[3]  
[Anonymous], 2004, ROBUST STAT
[4]   Robust inversion, dimensionality reduction, and randomized sampling [J].
Aravkin, Aleksandr ;
Friedlander, Michael P. ;
Herrmann, Felix J. ;
van Leeuwen, Tristan .
MATHEMATICAL PROGRAMMING, 2012, 134 (01) :101-125
[5]   ROBUST AND TREND-FOLLOWING STUDENT'S T KALMAN SMOOTHERS [J].
Aravkin, Aleksandr Y. ;
Burke, James V. ;
Pillonetto, Gianluigi .
SIAM JOURNAL ON CONTROL AND OPTIMIZATION, 2014, 52 (05) :2891-2916
[6]  
Aravkin AY, 2013, J MACH LEARN RES, V14, P2689
[7]   Estimating nuisance parameters in inverse problems [J].
Aravkin, Aleksandr Y. ;
van Leeuwen, Tristan .
INVERSE PROBLEMS, 2012, 28 (11)
[8]   An l1-Laplace Robust Kalman Smoother [J].
Aravkin, Aleksandr Y. ;
Bell, Bradley M. ;
Burke, James V. ;
Pillonetto, Gianluigi .
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 2011, 56 (12) :2892-2905
[9]   Hybrid l(1)/l(2) minimization with applications to tomography [J].
Bube, KP ;
Langan, RT .
GEOPHYSICS, 1997, 62 (04) :1183-1195
[10]   Doubly Robust Smoothing of Dynamical Processes via Outlier Sparsity Constraints [J].
Farahmand, Shahrokh ;
Giannakis, Georgios B. ;
Angelosante, Daniele .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2011, 59 (10) :4529-4543