Abstract

SARRIA-PAJA, M.  and  CASTELLANOS-DOMINGUEZ, G.. DISCRIMINATIVE TRAINING BASED ON MAHALANOBIS DISTANCE FOR PATHOLOGIC VOICE DETECTION. Dyna rev.fac.nac.minas [online]. 2010, vol.77, n.164, pp.220-228. ISSN 0012-7353.

This paper presents an approach that improves discriminative training criterion for Hidden Markov Models, and oriented to voice pathological identification. This technique aims at maximizing the Area under Curve of a Receiver Operating Characteristic curve by adjusting the model parameters using as objective function the Mahalanobis distance. The results show that the proposed technique outperforms significantly the accuracy in a classification system comparing with other training criteria. Results are provided using the MEEIVL voice disorders database.

Keywords : HMM; MLE; Discriminative training; voice pathology; ROC curve.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )