Genetic Algorithms as an Alternative Method of Parameter Estimation and Finding Most Likely Sequences of States of Hidden Markov Chains for HMMs and Hybrid HMM/ANN Models.
Katarzyna BijakPublished in: Fundam. Informaticae (2008)
Keyphrases
- parameter estimation
- hidden markov models
- markov chain
- em algorithm
- markov model
- genetic algorithm
- transition matrix
- neural network
- dynamic programming
- least squares
- continuous hidden markov models
- model selection
- maximum likelihood
- random walk
- artificial neural networks
- posterior distribution
- maximum likelihood estimation
- markov chain monte carlo
- transition probabilities
- markov random field
- support vector machine
- gibbs sampling
- probabilistic model
- parameter estimates
- markov processes
- expectation maximization
- markov models
- state space
- pairwise
- bayesian networks
- image processing
- data mining