ACL RD-TEC 1.0 Summarization of P04-1062
Paper Title:
ANNEALING TECHNIQUES FOR UNSUPERVISED STATISTICAL LANGUAGE LEARNING
ANNEALING TECHNIQUES FOR UNSUPERVISED STATISTICAL LANGUAGE LEARNING
Authors: Noah A. Smith and Jason Eisner
Primarily assigned technology terms:
- algorithm
- approximation
- classification
- classifier
- clustering
- collapsing
- compiler
- computing
- cross-validation
- discriminative training
- dp algorithm
- dynamic programming
- em algorithm
- expectation-maximization
- expectation-maximization algorithm
- forward pass
- forward-backward algorithm
- gradient-based method
- grammar induction
- hierarchical clustering
- hmms
- induction
- iterative scaling
- language and speech processing
- language learning
- learner
- learning
- learning method
- machine translation
- markov random fields
- matching
- maximum entropy
- maximum likelihood
- nlp
- optimization
- parameter estimation
- parameter search
- parameterization
- parsing
- part-of-speech tagging
- pos tagger
- pos tagging
- processing
- recognition
- search
- search algorithm
- simulated annealing
- smoothing
- speech processing
- speech recognition
- splitting
- statistical language learning
- structure-sharing
- supervised training
- tagger
- tagging
- training algorithm
- unsupervised learning
- unsupervised parameter estimation
- viterbi
Other assigned terms:
- approach
- bias
- binary tree
- case
- classification error
- classification error rate
- cluster
- clusters
- coefficient
- conditional distribution
- conditional model
- conditional probabilities
- conditional probability
- context features
- convergence
- corpora
- cross entropy
- cross-validation experiment
- derivation
- derivations
- dictionary
- distribution
- entropy
- entropy models
- error rate
- estimation
- events
- feature
- grammar
- grammatical categories
- implementation
- induction model
- interpolation
- knowledge
- language data
- language structure
- large corpora
- likelihood
- likelihood function
- local maxima
- local maximum
- log-likelihood
- maximum entropy models
- maximum entropy principle
- measures
- method
- names
- natural language
- nlp applications
- nouns
- parameter values
- parse
- parsing models
- part-of-speech
- part-of-speech tagging task
- pcfg
- penn treebank
- penn treebank corpus
- polynomial time
- pos tag
- posterior
- posterior distribution
- precision
- predictive power
- probabilities
- probability
- punctuation
- recognition accuracy
- runtime
- sentence
- sentences
- set size
- signal
- statistical grammar
- subcorpus
- symbols
- tag sequence
- tagging model
- tagging task
- tags
- term
- terms
- test corpus
- test data
- test set
- text
- theory
- tokens
- training
- training corpus
- training data
- training set
- training set size
- transition probabilities
- tree
- treebank
- treebank corpus
- trees
- trigram
- understanding
- uniform distribution
- unlabeled corpora
- unlabeled corpus
- unlabeled examples
- unlabeled text
- unlabeled training set
- word
- word sequence
- words
