EM-algorithm for training of state-space models with application to time series prediction.
Elia LiitiäinenNima ReyhaniAmaury LendassePublished in: ESANN (2006)
Keyphrases
- em algorithm
- expectation maximization
- parameter estimation
- state space
- mixture model
- maximum likelihood
- incomplete data
- maximum likelihood estimation
- probabilistic model
- hidden variables
- gaussian mixture model
- expectation maximisation
- baum welch
- structured prediction
- gaussian mixture
- generative model
- parameter learning
- finite mixture model
- density estimation
- probability density function
- likelihood function
- log likelihood
- search space
- log likelihood function
- update equations
- learning algorithm
- probabilistic principal component analysis
- mixture modeling
- model based clustering
- unsupervised learning
- model selection
- reinforcement learning