Recovery of nested supports by greedy sparse representation algorithms over non-normalized dictionaries
Mailhe, B, Sturm, B and Plumbley, MD (2013) Recovery of nested supports by greedy sparse representation algorithms over non-normalized dictionaries In: In Proceedings of the Workshop on Signal Processing with Adaptive Sparse Structured Representations (SPARS 2013), 8-11 July 2013. [Abstract only], 2013-07-08-2013-07-11, Lausanne, Switzerland.
Available under License : See the attached licence file.
Download (33kB) | Preview
We prove that if Orthogonal Matching Pursuit (OMP) recovers all s-sparse signals for a given dictionary, then it also recovers all s 0 -sparse signals on the same dictionary for any s 0 < s. We also extend Tropp’s Exact Recovery Condition (ERC) to dictionaries with non-normalized atoms. Our result does not contradict an earlier result stating that there are dictionaries and cardinalities s 0 < s such that all s-size supports satisfy Tropp’s (ERC) but not all s 0 -size supports do: that result was proved using non-normalized dictionaries and in that case Tropp’s ERC is not linked to the recovery by OMP.
|Item Type:||Conference or Workshop Item (Conference Paper)|
|Related URLs :|
|Additional Information :||Full text not available from this repository.|
|Depositing User :||Symplectic Elements|
|Date Deposited :||08 Dec 2015 16:38|
|Last Modified :||08 Dec 2015 16:38|
Actions (login required)
Downloads per month over past year