Conflict intensity estimation from speech using Greedy forward-backward feature selection.
Gábor GosztolyaPublished in: INTERSPEECH (2015)
Keyphrases
- forward backward
- feature selection
- forward selection
- hidden markov models
- mutual information
- speech recognition
- text classification
- support vector
- partial volume
- text categorization
- support vector machine
- estimation accuracy
- feature set
- conflict resolution
- greedy algorithm
- broadcast news
- feature extraction
- machine learning
- irrelevant features
- method for feature selection
- selection criterion
- speech synthesis
- selected features
- intensity images
- feature selection algorithms
- feature space
- parameter estimation
- dynamic programming
- classification accuracy
- multi class
- simulated annealing
- decision trees
- recognition engine
- search space
- spoken language
- image intensity
- noisy environments
- model selection
- density estimation
- classification models
- information gain
- speech signal