ACL RD-TEC 1.0 Summarization of N04-4006
Paper Title:
LANGUAGE MODEL ADAPTATION WITH MAP ESTIMATION AND THE PERCEPTRON ALGORITHM
LANGUAGE MODEL ADAPTATION WITH MAP ESTIMATION AND THE PERCEPTRON ALGORITHM
Authors: Michiel Bacchiani and Brian Roark and Murat Saraclar
Primarily assigned technology terms:
- acoustic model adaptation
- algorithm
- automaton
- channel normalization
- constrained model-space adaptation
- decoding
- discriminant analysis
- discriminative modeling
- finite state
- finite state automaton
- iterative training
- language model adaptation
- lattice construction
- length normalization
- linear discriminant
- linear discriminant analysis
- linear regression
- lm adaptation
- map adaptation
- map estimation
- matching
- maximum likelihood
- measuring
- model adaptation
- model training
- modeling
- multi-pass recognition
- multi-pass rescoring
- n-gram model training
- normalization
- perceptron
- perceptron algorithm
- perceptron training
- recognition
- recognition process
- recognition system
- recognizer
- regression
- rescoring
- search
- speech recognition
- speech recognition system
- state automaton
- transcription
- voicemail
- voicemail transcription
Other assigned terms:
- acoustic model
- approach
- backoff
- backoff model
- bigram
- case
- customer care
- data set
- error rate
- estimation
- experimental results
- feature
- feature vector
- feature weight
- feature weights
- grammar
- hypotheses
- hypothesis
- language model
- lattice
- lattices
- lexicon
- likelihood
- linear model
- method
- n-gram
- n-gram model
- out-of-domain corpus
- probabilities
- process
- terms
- test data
- test set
- training
- training corpus
- training data
- training set
- transcripts
- trigram
- unigram
- utterance
- vocabulary
- vocal tract
- weight vector
- word
- word error rate
- word lattice
- word lattices
- words
