ACL RD-TEC 1.0 Summarization of N04-2010
Paper Title:
SPEAKER RECOGNITION WITH MIXTURES OF GAUSSIANS WITH SPARSE REGRESSION MATRICES
SPEAKER RECOGNITION WITH MIXTURES OF GAUSSIANS WITH SPARSE REGRESSION MATRICES
Primarily assigned technology terms:
- algorithm
- classification
- database
- decomposition
- discriminative parameter estimation
- em algorithm
- full regression
- generative parameter estimation
- graphical model
- greedy pruning
- identification
- iterative maximization
- learning
- likelihood estimation
- linear regression
- maximum likelihood
- maximum likelihood estimation
- modeling
- parameter adaptation
- parameter estimation
- parameter search
- pruning
- recognition
- regression
- scoring
- search
- searching
- speaker adaptation
- speaker identification
- speaker recognition
- speech recognition
- structure finding
- structure learning
- structure search
- structure-learning
Other assigned terms:
- approach
- case
- classification tasks
- coefficient
- dimensionality
- distribution
- error rate
- estimation
- fact
- feature
- implementation
- index
- likelihood
- likelihood parameter
- linear algebra
- log-likelihood
- mechanisms
- methodology
- modeling power
- mutual information
- nist
- np-hard problem
- parameter values
- pauses
- probability
- probability distribution
- recognition task
- sentence
- sentences
- statistics
- terms
- test data
- test set
- testing data
- training
- training data
- transformation
- trees
