Backoff inspired features for maximum entropy language models.
Fadi BiadsyKeith B. HallPedro J. MorenoBrian RoarkPublished in: INTERSPEECH (2014)
Keyphrases
- maximum entropy
- language model
- maximum entropy model
- language modeling
- n gram
- document retrieval
- maximum entropy principle
- test collection
- retrieval model
- probabilistic model
- transformation based learning
- language modelling
- principle of maximum entropy
- statistical language models
- smoothing methods
- named entity recognition
- query terms
- mixture model
- query expansion
- active learning
- feature vectors
- feature space
- language models for information retrieval
- information retrieval
- speech recognition
- text categorization
- higher order
- classification accuracy