Abstract | ||
---|---|---|
Using the framework of Reproducing Kernel Hilbert Spaces, we develop a new sequence kernel that mea- sures similarity between sequences of observations. We then apply it to a text-independent speaker verification task using the NIST 2004 Speaker Recognition Eval- uation database. The results show that incorporating our new sequence kernel in an SVM training architec- ture not only yields performance significantly superior to those of a baseline UBM-GMM classifier but also out- performs the Generalized Linear Discriminant Sequence (GLDS) Kernel classifier. Moreover, our kernel maps to a relatively low dimensional feature space while allowing a large choice for the kernel function. |
Year | Venue | Keywords |
---|---|---|
2005 | EUSIPCO | gaussian processes,hilbert spaces,mixture models,speaker recognition,support vector machines,gaussian mixture models,nist 2004 speaker recognition evaluation database,svm speaker verification,svm training architecture,ubm-gmm classifier,kernel hilbert spaces,low dimensional feature space,sequence kernel,text-independent speaker verification task |
Field | DocType | ISBN |
Graph kernel,Radial basis function kernel,Pattern recognition,Kernel embedding of distributions,Computer science,Speech recognition,Tree kernel,Polynomial kernel,Artificial intelligence,Kernel method,String kernel,Kernel (statistics) | Conference | 978-160-4238-21-1 |
Citations | PageRank | References |
4 | 0.49 | 9 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jérôme Louradour | 1 | 829 | 55.81 |
Khalid Daoudi | 2 | 145 | 23.68 |