Penalized Logistic Regression With HMM Log-Likelihood Regressors for Speech Recognition.
Øystein BirkenesTomoko MatsuiKunio TanabeSabato Marco SiniscalchiTor André MyrvollMagne Hallstein JohnsenPublished in: IEEE Trans. Speech Audio Process. (2010)
Keyphrases
- logistic regression
- log likelihood
- speech recognition
- maximum likelihood
- hidden markov models
- loss function
- speaker adaptation
- information theoretic
- em algorithm
- speech signal
- decision trees
- automatic speech recognition
- density estimation
- language model
- support vector
- speaker independent
- exponential family
- speech recognizer
- naive bayes
- scoring function
- least squares
- pattern recognition
- speech recognition systems
- neural network
- closed form
- computer vision
- pairwise
- mutual information
- markov networks
- expectation maximization
- supervised learning