Basis Identification from Random Sparse SamplesReport as inadecuate

Basis Identification from Random Sparse Samples - Download this document for free, or read online. Document in PDF available to download.

Presented at: SPARS09, Saint-Malo, April 6-9, 2009 Published in: Proc. SPARS09 Saint-Malo: , 2009

This article treats the problem of learning a dictionary providing sparse representations for a given signal class, via $\ell_1$ minimisation. The problem is to identify a dictionary $\dico$ from a set of training samples $\Y$ knowing that $\Y = \dico \X$ for some coefficient matrix $\X$. Using a characterisation of coefficient matrices $\X$ that allow to recover any basis as a local minimum of an $\ell_1$ minimisation problem, it is shown that certain types of sparse random coefficient matrices will ensure local identifiability of the basis with high probability. The necessary number of training samples grows up to a logarithmic factor linearly with the signal dimension.

Keywords: basis identification ; $\ell_1$ minimisation ; sparse samples ; LTS2 ; lts2 Reference EPFL-CONF-134108

Author: Gribonval, Remi; Schnass, Karin


Related documents