Global and componentwise extrapolations for accelerating training of Bayesian networks and conditional random fields.
Han-Shen HuangBo-Hou YangYu-Ming ChangChun-Nan HsuPublished in: Data Min. Knowl. Discov. (2009)
Keyphrases
- conditional random fields
- bayesian networks
- structured prediction
- graphical models
- parameter learning
- probabilistic model
- supervised training
- hidden markov models
- exact inference
- sequence labeling
- information extraction
- random fields
- higher order
- restricted boltzmann machine
- generative model
- maximum entropy
- error correcting output coding
- markov random field
- semi markov
- markov networks
- crf model
- probabilistic graphical models
- named entity recognition
- approximate inference
- web page prediction
- structure learning
- structured learning
- probabilistic inference
- protein fold recognition
- pairwise
- training set
- weighted sums
- conditional independence
- conditional probabilities
- belief propagation
- training samples
- undirected graphical models
- supervised learning
- image processing
- latent variables
- multi class
- support vector
- decision trees