DocumentCode :
2057273
Title :
Matrix completion algorithms with optimal phase transition
Author :
Tanner, Jared ; Ke Wei
Author_Institution :
Math. Inst., Univ. of Oxford, Oxford, UK
fYear :
2013
fDate :
9-13 Sept. 2013
Firstpage :
1
Lastpage :
2
Abstract :
We present new first order algorithm for matrix completion and show by numerical tests that they are able to recover rank r matrices of size m × n from only Const. · (m + n - r)r entries with Const. close to one. This evidence shows that matrix completion is achieved closer to the oracle limit than in compressed sensing.
Keywords :
compressed sensing; concave programming; matrix algebra; compressed sensing; first order algorithm; matrix completion algorithms; nonconvex optimization; optimal phase transition; oracle limit; Abstracts; Matrix Completion; Nonconvex Optimization;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Signal Processing Conference (EUSIPCO), 2013 Proceedings of the 21st European
Conference_Location :
Marrakech
Type :
conf
Filename :
6811586
Link To Document :
بازگشت