A GIS-like training algorithm for log-linear models with hidden variables.
Georg HeigoldThomas DeselaersRalf SchlüterHermann NeyPublished in: ICASSP (2008)
Keyphrases
- training algorithm
- log linear
- models with hidden variables
- hidden variables
- latent variables
- graphical models
- back propagation
- neural network
- probabilistic modeling
- log linear models
- hidden layer
- probabilistic model
- training process
- support vector machine
- discriminative training
- efficient learning
- learning algorithm
- bayesian networks
- artificial neural networks
- mixture model
- rbf neural network
- learning rate
- generative model
- prior knowledge
- random variables
- approximate inference
- fuzzy logic
- state space
- training data
- machine learning