Sparsity Measure and the Detection of Significant Data
Résumé
The paper provides a formal description of the sparsity of a representation via the detection thresholds. The formalism proposed derives from theoretical results about the detection of significant coefficients when data are observed in presence of additive white Gaussian noise. The detection thresholds depend on two parameters describing the sparsity degree for the representation of a signal. The standard universal and minimax thresholds correspond to detection thresholds associated with different sparsity degrees.
Origine | Fichiers produits par l'(les) auteur(s) |
---|