Error - weighted discriminative training for HMM parameter estimation.
Daniel WillettPublished in: INTERSPEECH (2004)
Keyphrases
- parameter estimation
- discriminative training
- maximum likelihood
- hidden markov models
- em algorithm
- statistical models
- maximum likelihood estimation
- model selection
- minimum classification error
- least squares
- probabilistic interpretation
- parameter estimation algorithm
- speech recognition
- markov random field
- automatic speech recognition
- expectation maximization
- gaussian mixture model
- posterior probability
- maximum a posteriori
- image reconstruction
- gibbs sampling
- data mining