EM Algorithm and -ICA Learning Based upon Extended Logarithmic Information Measures.
Yasuo MatsuyamaTakeshi NiimotoNaoto KatsumataYoshitaka SuzukiSatoshi FurukawaPublished in: IJCNN (3) (2000)
Keyphrases
- em algorithm
- expectation maximization
- parameter learning
- hidden variables
- prior knowledge
- generative model
- gaussian mixture
- maximum likelihood estimation
- mixture model
- parameter estimation
- maximum likelihood
- feature extraction
- likelihood maximization
- incomplete data
- likelihood function
- log likelihood
- mixture modeling
- unsupervised learning
- density estimation
- learning algorithm
- supervised learning
- active learning
- pairwise
- training data
- face recognition