Entrenamiento Discriminativo Maximizando una Distancia entre Modelos de Clases

Abstract

This paper presents an approach that improves discriminative training criterion for Hidden Markov Models, and it is oriented to voice pathological identification. This technique aims at maximizing the Area under the Receiver Operating Characteristic curve by adjusting the model parameters using as objective function the distance between the means of the underlying probability densities functions associated with each class. As result we obtain an improvement in the performance of the classification system compared with different training criteria.