Regularized EM algorithm for analysis of incomplete datasets
A regularized expectation-maximization (EM) algorithm for the estimation of mean values and covariance matrices and for the imputation of missing values in large, incomplete datasets.