Training of lexical models based on DTW-based parameter reestimation algorithm.
Yoshiharu AbeKunio NakajimaPublished in: ICASSP (1988)
Keyphrases
- experimental evaluation
- search space
- probabilistic model
- cost function
- times faster
- optimization algorithm
- recognition algorithm
- dynamic programming
- detection algorithm
- worst case
- preprocessing
- search algorithm
- k means
- objective function
- np hard
- learning algorithm
- simulated annealing
- neural network
- parameter estimation
- linear model
- bayesian framework
- training process
- convergence rate
- matching algorithm
- model selection
- significant improvement
- computational complexity
- training data
- genetic algorithm