An EM algorithm for learning sparse and overcomplete representations.
Mingjun ZhongHuanwen TangHongjun ChenYiyuan TangPublished in: Neurocomputing (2004)
Keyphrases
- em algorithm
- hidden variables
- maximum likelihood
- mixture model
- expectation maximization
- gaussian mixture model
- learning algorithm
- generative model
- reinforcement learning
- active learning
- parameter estimation
- sparse representation
- prior knowledge
- incomplete data
- parameter learning
- closed form
- log likelihood
- gaussian mixture
- maximum likelihood estimation
- maximum a posteriori
- discriminative learning
- mixture distribution
- likelihood maximization
- finite mixtures