On the EM algorithm for the estimation of speech AR parameters in noise.
Marcin KuropatwinskiW. Bastiaan KleijnPublished in: ICASSP (2014)
Keyphrases
- em algorithm
- maximum likelihood estimation
- expectation maximization
- parameter estimation
- maximum likelihood
- log likelihood function
- update equations
- likelihood function
- mixture model
- expectation maximisation
- maximum likelihood estimates
- gaussian mixture
- hyperparameters
- gaussian mixture model
- probabilistic model
- density estimation
- generative model
- maximum a posteriori
- parameter learning
- log likelihood
- map estimation
- incomplete data
- unsupervised learning
- mixture modeling
- mixture of gaussians
- finite mixture model
- probability density function
- hidden variables
- free energy
- graphical models
- mixture distribution
- estimate the model parameters
- penalized likelihood
- image processing
- noise reduction
- bayesian framework
- missing data
- model selection
- multi modal
- pairwise