Comparison of a gradient-based and LASSO (ISTA) algorithm for sparse signal reconstruction

Sparse signal reconstruction performed by two different algorithms is considered. First algorithm is the ISTA algorithm for LASSO minimization, while the second one is the gradient-based descent algorithm. Algorithms perform signal reconstruction in a completely different way. The ISTA algorithm rec...

Full description

Saved in:
Bibliographic Details
Published in2016 5th Mediterranean Conference on Embedded Computing (MECO) pp. 377 - 380
Main Authors Vujovic, Stefan, Stankovic, Isidora, Dakovic, Milos, Stankovic, Ljubisa
Format Conference Proceeding
LanguageEnglish
Published IEEE 28.07.2016
Subjects
Bars
Compressed sensing
Compressive sensing
Discrete cosine transforms
Embedded computing
Gradient algorithm
Ista algorithm
Iterative algorithms
Lasso minimization
Minimization
Signal processing algorithms
Signal reconstruction
Sparse reconstruction
Sparse signals
Transforms
Vectors
Online AccessGet full text
DOI10.1109/MECO.2016.7525785

Cover

More Information
Summary:Sparse signal reconstruction performed by two different algorithms is considered. First algorithm is the ISTA algorithm for LASSO minimization, while the second one is the gradient-based descent algorithm. Algorithms perform signal reconstruction in a completely different way. The ISTA algorithm reconstructs signals in the sparsity transformation domain. The gradient descent algorithm performs reconstruction in time/measurements domain, considering the missing samples as variables. Both of them use the l1-norm in minimization. Computational time and mean absolute error are used in comparison analysis presented in this paper.
DOI:10.1109/MECO.2016.7525785