000150577 001__ 150577
000150577 005__ 20190316234842.0
000150577 037__ $$aREP_WORK
000150577 245__ $$aInvestigation of kNN Classifier on Posterior Features Towards Application in Automatic Speech Recognition
000150577 269__ $$a2010
000150577 260__ $$bIdiap$$c2010
000150577 336__ $$aReports
000150577 520__ $$aClass posterior distributions can be used to classify or as intermediate features, which can be further exploited in different classifiers (e.g., Gaussian Mixture Models, GMM) towards improving speech recognition performance. In this paper we examine the possibility to use kNN classifier to perform local phonetic classification of class posterior distribution extracted from acoustic vectors. In that framework, we also propose and evaluate a new kNN metric based on the relative angle between feature vectors to define the nearest neighbors. This idea is inspired by the orthogonality characteristic of the posterior features. To fully exploit this attribute, kNN is used in two main steps: (1) the distance is computed as the cosine function of the relative angle between the test vector and the training vector and (2) the nearest neighbors are defined as the samples within a specific relative angle to the test data and the test samples which do not have enough labels in such a hyper-cone are considered as uncertainties and left undecided. This approach is evaluated on TIMIT database and compared to other metrics already used in literature for measuring the similarity between posterior probabilities. Based on our experiments, the proposed approach yield 78.48% frame level accuracy while specifying 15.17% uncertainties in the feature space.
000150577 700__ $$0243353$$g188259$$aAsaei, Afsaneh
000150577 700__ $$g117014$$aBourlard, Hervé$$0243348
000150577 700__ $$aPicart, Benjamin
000150577 8564_ $$uhttp://publications.idiap.ch/downloads/reports/2009/Asaei_Idiap-RR-11-2010.pdf$$zURL
000150577 8564_ $$uhttps://infoscience.epfl.ch/record/150577/files/Asaei_Idiap-RR-11-2010.pdf$$zn/a$$s896415
000150577 909C0 $$xU10381$$0252189$$pLIDIAP
000150577 909CO $$ooai:infoscience.tind.io:150577$$qGLOBAL_SET$$pSTI$$preport
000150577 937__ $$aEPFL-REPORT-150577
000150577 970__ $$aAsaei_Idiap-RR-11-2010/LIDIAP
000150577 973__ $$sPUBLISHED$$aEPFL
000150577 980__ $$aREPORT