Error type classification and word accuracy estimation using alignment features from word confusion network.
Atsunori OgawaTakaaki HoriAtsushi NakamuraPublished in: ICASSP (2012)
Keyphrases
- classification accuracy
- co occurrence
- feature set
- feature extraction
- error analysis
- feature vectors
- feature space
- feature reduction
- n gram
- error rate
- lexical features
- extracted features
- classification process
- classification method
- classification performances
- class labels
- multiple types
- roc curve
- feature subset
- estimation error
- word level
- machine learning
- feature selection
- relative error
- word pairs
- support vector machine classifier
- punctuation marks
- decision trees
- svm classifier
- training set
- word sense disambiguation
- training data
- handwritten words
- support vector machine
- linguistic information
- naive bayes
- benchmark datasets
- classification rate
- classification models
- text classification
- image classification
- gender classification
- estimation accuracy
- image features
- feature values
- word alignment
- discriminatory power
- support vector