ACL RD-TEC 1.0 Summarization of P05-1044
Paper Title:
CONTRASTIVE ESTIMATION: TRAINING LOG-LINEAR MODELS ON UNLABELED DATA
CONTRASTIVE ESTIMATION: TRAINING LOG-LINEAR MODELS ON UNLABELED DATA
Authors: Noah A. Smith and Jason Eisner
Primarily assigned technology terms:
- algorithm
- approximation
- automaton
- bootstrapping
- capitalization
- classification
- computing
- conditional estimation
- conditional random fields
- contrastive estimation
- crfs
- dependency grammar induction
- dynamic programming
- dynamic programming algorithm
- em algorithm
- error-correction
- expectation-maximization
- finite-state automaton
- forward-backward algorithm
- grammar induction
- hmms
- induction
- lattice parsing
- learner
- learning
- learning natural language
- margin training
- markov random field
- maximum entropy
- model selection
- modeling
- namedentity extraction
- normalization
- numerical optimization
- numerical optimization method
- optimization
- optimization method
- parameter estimation
- parsing
- pos tagging
- probabilistic modeling
- programming algorithm
- random sampling
- regularization
- sampling
- search
- semi-supervised classification
- sequence labeling
- sequence modeling
- shallow parsing
- smoothing
- smoothing techniques
- spelling
- spelling correction
- statistical learning
- subcategorization
- tagger
- tagging
- transducers
- trigram tagger
- tuning
- unsupervised estimation
- unsupervised learning
- unsupervised sequence modeling
- viterbi
Other assigned terms:
- acoustic models
- alphabet
- ambiguous words
- approach
- bias
- case
- classification problem
- composition
- convergence
- correlations
- dependency grammar
- development set
- dictionaries
- dictionary
- distribution
- domain knowledge
- entropy
- estimation
- feature
- feature set
- feature sets
- feature types
- feature value
- feature weight
- feature weights
- gaussian prior
- generative model
- generative models
- grammar
- hypotheses
- hypothesis
- improved search
- input string
- kl divergence
- knowledge
- labeling
- language syntax
- lattice
- lattices
- lexemes
- lexicon
- likelihood
- linguistic
- linguistic structure
- linguists
- local maxima
- local maximum
- log-linear models
- mapping
- maximum entropy principle
- meaning
- measure
- method
- namedentity
- natural language
- natural language syntax
- oracle
- pcfgs
- pos sequence
- pos tag
- probabilistic model
- probabilistic models
- probability
- punctuation
- relation
- representations
- sentence
- sentences
- smoothing parameter
- substring
- suffixes
- symbol
- syntactic model
- syntax
- tag set
- tagging accuracy
- tags
- technique
- term
- terms
- text
- topology
- training
- training data
- training example
- training examples
- training set
- trigram
- unigram
- unigram model
- unlabeled examples
- unlabeled text
- vocabulary
- word
- word order
- word types
- words
