Hybrid generative-discriminative training of Gaussian mixture models

Computer Science

• Gaussian Mixture Model 100%
• maximum-likelihood 33%
• Good Performance 33%
• Social Exclusion 33%
• Performance Improvement 33%
• Unlabeled Data 33%
• Covariance Matrix 33%
• Diagonal Matrix 33%
• Stochastic Optimization 33%
• Structure Matrix 33%
• Conditional Likelihood 33%
• Input Distribution 33%