Stochastic Gradient Variational Bayes for deep learning-based ASR.
Andros TjandraSakriani SaktiSatoshi NakamuraMirna AdrianiPublished in: ASRU (2015)
Keyphrases
- deep learning
- hyperparameters
- unsupervised learning
- posterior distribution
- bayesian inference
- machine learning
- model selection
- latent variables
- bayesian framework
- log likelihood
- cross validation
- closed form
- gaussian mixture model
- random sampling
- data sets
- natural images
- support vector
- prior information
- markov chain monte carlo
- image processing
- incomplete data
- pattern recognition
- maximum likelihood
- maximum a posteriori
- high dimensional
- em algorithm
- expectation maximization