Greedy rank updates combined with Riemannian descent methods for low-rank optimization

被引:0
作者
Uschmajew, Andre [1 ,2 ]
Vandereycken, Bart [3 ]
机构
[1] Univ Bonn, Hausdorff Ctr Math, D-53115 Bonn, Germany
[2] Univ Bonn, Inst Numer Simulat, D-53115 Bonn, Germany
[3] Univ Geneva, Sect Math, CH-1211 Geneva, Switzerland
来源
2015 INTERNATIONAL CONFERENCE ON SAMPLING THEORY AND APPLICATIONS (SAMPTA) | 2015年
关键词
MATRIX COMPLETION; DECOMPOSITIONS;
D O I
暂无
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
We present a rank-adaptive optimization strategy for finding low-rank solutions of matrix optimization problems involving a quadratic objective function. The algorithm combines a greedy outer iteration that increases the rank and a smooth Riemannian algorithm that further optimizes the cost function on a fixed-rank manifold. While such a strategy is not especially novel, we show that it can be interpreted as a perturbed gradient descent algorithms or as a simple warm-starting strategy of a projected gradient algorithm on the variety of matrices of bounded rank. In addition, our numerical experiments show that the strategy is very efficient for recovering full rank but highly ill-conditioned matrices that have small numerical rank.
引用
收藏
页码:420 / 424
页数:5
相关论文
共 23 条
[1]   ACCELERATED LINE-SEARCH AND TRUST-REGION METHODS [J].
Absil, P. -A. ;
Gallivan, K. A. .
SIAM JOURNAL ON NUMERICAL ANALYSIS, 2009, 47 (02) :997-1018
[2]  
Absil PA, 2008, OPTIMIZATION ALGORITHMS ON MATRIX MANIFOLDS, P1
[3]  
[Anonymous], 2014, Propack-software for large and sparse svd calculations
[4]   Exact Matrix Completion via Convex Optimization [J].
Candes, Emmanuel J. ;
Recht, Benjamin .
FOUNDATIONS OF COMPUTATIONAL MATHEMATICS, 2009, 9 (06) :717-772
[5]   Iterative methods for low rank approximation of graph similarity matrices [J].
Cason, T. P. ;
Absil, P. -A. ;
Van Dooren, P. .
LINEAR ALGEBRA AND ITS APPLICATIONS, 2013, 438 (04) :1863-1882
[6]   On smooth decompositions of matrices [J].
Dieci, L ;
Eirola, T .
SIAM JOURNAL ON MATRIX ANALYSIS AND APPLICATIONS, 1999, 20 (03) :800-819
[7]  
Dolgov S. V., 2013, ARXIV13016068
[8]   Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions [J].
Halko, N. ;
Martinsson, P. G. ;
Tropp, J. A. .
SIAM REVIEW, 2011, 53 (02) :217-288
[9]  
Hammarling S., 2008, UPDATING QR FACTORIZ
[10]   CRITICAL-POINTS OF MATRIX LEAST-SQUARES DISTANCE FUNCTIONS [J].
HELMKE, U ;
SHAYMAN, MA .
LINEAR ALGEBRA AND ITS APPLICATIONS, 1995, 215 :1-19