DocumentCode :
2981472
Title :
Efficient and guaranteed rank minimization by atomic decomposition
Author :
Lee, Kiryung ; Bresler, Yoram
Author_Institution :
Dept. of ECE, Univ. of Illinois at Urbana-Champaign, Urban, IL, USA
fYear :
2009
fDate :
June 28 2009-July 3 2009
Firstpage :
314
Lastpage :
318
Abstract :
Recht, Fazel, and Parrilo provided an analogy between rank minimization and ¿0-norm minimization. Subject to the rank-restricted isometry property, nuclear norm minimization is a guaranteed algorithm for rank minimization. The resulting semidefinite formulation is a convex problem but in practice the algorithms for it do not scale well to large instances. Instead, we explore missing terms in the analogy and propose a new algorithm which is computationally efficient and also has a performance guarantee. The algorithm is based on the atomic decomposition of the matrix variable and extends the idea in the CoSaMP algorithm for ¿0-norm minimization. Combined with the recent fast low rank approximation of matrices based on randomization, the proposed algorithm can efficiently handle large scale rank minimization problems.
Keywords :
matrix algebra; minimisation; CoSaMP algorithm; atomic decomposition; convex problem; matrix variable; missing terms; nuclear norm minimization; performance guarantee; rank minimization; rank-restricted isometry property; ¿0-norm minimization; Approximation algorithms; Compressed sensing; Focusing; Image coding; Iterative algorithms; Large-scale systems; Linear systems; Matching pursuit algorithms; Matrix decomposition; Minimization methods;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Information Theory, 2009. ISIT 2009. IEEE International Symposium on
Conference_Location :
Seoul
Print_ISBN :
978-1-4244-4312-3
Electronic_ISBN :
978-1-4244-4313-0
Type :
conf
DOI :
10.1109/ISIT.2009.5205530
Filename :
5205530
Link To Document :
بازگشت