How to select a good training-data subset for transcription: submodular active selection for sequences.
Hui LinJeff A. BilmesPublished in: INTERSPEECH (2009)
Keyphrases
- training data
- selection algorithm
- subset selection
- training set
- test data
- dynamically select
- hidden markov models
- data sets
- decision trees
- automatic selection
- training samples
- training process
- feature subset
- greedy algorithm
- high order
- training examples
- supervised learning
- sequence alignment
- long sequences
- learning algorithm
- handwriting recognition
- classification models
- class labels
- noisy data
- sequential patterns
- unlabeled data
- small number
- classification accuracy
- domain knowledge
- prior knowledge
- objective function
- genetic algorithm