A symmetrization of the Subspace Gaussian Mixture Model.
Daniel PoveyMartin KarafiátArnab GhoshalPetr SchwarzPublished in: ICASSP (2011)
Keyphrases
- gaussian mixture model
- feature space
- covariance matrices
- mixture model
- feature vectors
- maximum likelihood
- principal component analysis
- gaussian mixture
- low dimensional
- speaker recognition
- background subtraction
- em algorithm
- high dimensional
- expectation maximization
- dimensionality reduction
- speaker identification
- high dimensional data
- probability density function
- gaussian distribution
- bayesian information criterion
- principal components
- image set
- kernel pca
- prior knowledge
- multiscale
- computer vision