: Vocabulario de PLN: parameter estimation

NLP methods and tools > parameter estimation

parameter estimation

In order to reach reliable parameter estimations for so many features discriminative training for MT must be scaled to large amount of training data. (Cao & Khudanpur, 2012)
The experiments showed that this approach obtains significantly better results than using the maximum mutual information criterion on parameter estimation. (Shen, Sarkar & Och, 2004)
This alignment however is computed by using the parameter estimations of models that are only a (raw) approximation of the true parameter distributions but are treated as an additional sure knowledge source. (Vilar, Popovic & Ney, 2006)
This is a loosely coupled model since parameter estimation is performed separately for the two sub-models. (Yu & Xie, 2014)

http://data.loterre.fr/ark:/67375/8LP-GVRK2NVC-3

RDF/XML TURTLE JSON-LD última modificación 10/6/24

Loterre