Unsupervised Selection and Discriminative Estimation of Orthogonal Gaussian Mixture Models for Handwritten Digit Recognition.
Xuefeng ChenXiabi LiuYunde JiaPublished in: ICDAR (2009)
Keyphrases
- gaussian mixture model
- handwritten digit recognition
- mixture model
- probability density
- unsupervised learning
- density estimation
- semi supervised
- maximum likelihood criterion
- expectation maximization
- em algorithm
- feature vectors
- speaker recognition
- maximum likelihood
- feature space
- speaker identification
- log likelihood function
- covariance matrices
- probability density function
- parameter estimation
- gaussian mixture
- feature extraction
- variational bayes
- supervised learning
- feature selection
- maximum likelihood estimation
- gaussian distribution
- computer vision
- finite mixtures