Quasi-Newton method for maximum likelihood estimation of hidden Markov models.
Olivier CappéVincent BuchouxEric MoulinesPublished in: ICASSP (1998)
Keyphrases
- hidden markov models
- maximum likelihood estimation
- quasi newton method
- objective function
- em algorithm
- limited memory
- quasi newton
- maximum likelihood
- parameter estimation
- expectation maximization
- probability distribution
- density function
- conditional random fields
- discriminative training
- minimum classification error
- machine learning