A unified approach on fast training of feedforward and recurrent networks using EM algorithm.
Sheng MaChuanyi JiPublished in: IEEE Trans. Signal Process. (1998)
Keyphrases
- recurrent networks
- em algorithm
- feed forward
- expectation maximization
- recurrent neural networks
- mixture model
- maximum likelihood
- neural network
- artificial neural networks
- back propagation
- gaussian mixture model
- parameter estimation
- generative model
- maximum likelihood estimation
- likelihood function
- incomplete data
- expectation maximisation
- error back propagation
- hidden layer
- log likelihood
- gaussian mixture
- visual cortex
- biologically inspired
- hidden variables
- maximum a posteriori
- penalized likelihood
- adaptive neural
- mixture modeling
- probability density function
- artificial intelligence
- higher order
- finite mixture model
- probabilistic model
- feature vectors
- unsupervised learning
- model based clustering
- fuzzy logic
- object recognition