A Framework for pre-training hidden-unit conditional random fields and its extension to long short term memory networks.
Young-Bum KimKarl StratosRuhi SarikayaPublished in: Comput. Speech Lang. (2017)
Keyphrases
- conditional random fields
- probabilistic model
- named entity recognition
- hidden markov models
- error correcting output coding
- training set
- fully connected
- structured prediction
- graphical models
- maximum entropy
- active learning
- key features
- neural network
- random fields
- markov random field
- max margin
- pairwise
- crf model
- discriminative models
- structured learning
- sequence labeling
- supervised training
- data mining