The EM algorithm and related statistical models

摘要: Incomplete data and the generation mechanisms type of incomplete its analysis statistical models for with missing values in multinominal algorithms MLE multivariate normal datawith scoring method EM algorithm basics extension acceleration as an optimization tool robust model outlier detection scale mixture distributions andcontaminated distribution tobit factor latent variables structure class trait structured equations extensions ECM ECME optimal MCEM covergence speed convergence comparisons other quasi Newton methods EMalgorithm neural networks geometric interpretation Marcov chain Monte Carlo Bayes estimation Metropolis-Hastings augmentation poor man's dataaugmentation Gibbs sampling algorithm. Appendices: SOLAS Lem.