Constructive approximations to the q = 1/2 maximum entropy distribution from redundant and noisy data

The problem of constructing the q = 1/2 nonextensive maximum entropy distributions from redundant and noisy data is considered. A strategy is proposed, which evolves through the following steps. (i) Independent constraints are first preselected by recourse to a data-independent technique to be discu...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Rebollo Neira, Laura, Plastino, Ángel Luis
Formato: Articulo
Lenguaje:Inglés
Publicado: 2004
Materias:
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/126470
Aporte de:
Descripción
Sumario:The problem of constructing the q = 1/2 nonextensive maximum entropy distributions from redundant and noisy data is considered. A strategy is proposed, which evolves through the following steps. (i) Independent constraints are first preselected by recourse to a data-independent technique to be discussed here. (ii) The data are <i>a posteriori</i> used to determine the parameters of the distribution by a previously introduced forward approach. (iii) A backward approach is proposed for reducing the parameters of such distribution. The previously introduced forward approach is generalized here in order to make it suitable for dealing with very noisy data.