gemEstimator: Estimates the Gaussian mixture parameters using GEM algorithm
In km20/gemalgorithm: Graphical Expectation Maximization

This function estimates the Gaussian mixture parameters using the GEM algorithm. The mixture components number is supposed to be known. This function uses an initial parameters guess and a set of associated graphs to iteratively estimate the parameters.

1	gemEstimator(x, Nit, p0, mu0, sigma0, G = list())

`x`	A matrix of observations with d rows (number of variables) and n columns (number of observations).
`Nit`	An integer that specifies the iterations number. If a negative value is specified then a stopping rule is used.
`p0`	A numeric vector of initial proportions.
`mu0`	A list of numeric vectors which are the initial mean of each mixture component.
`sigma0`	A list of covariance matrices (Algorithm initialization)
`G`	A list of K graphs associated with the mixture components. Each graph is a named list with three items : Cliques list, Sperators list and Sperators multiplicities. If an empty list is provided then the classical EM algorithm is used.

This function returns a named list containing the following results: 'p': A vector of estimated proportions. 'mu': A list of estimated mean vectors. 'sigma' : A list of estimated covarianece matrices which takes into account the provided graph structures. 'tau' : An n by K matrix that contains the posterior probabilities that each observation belongs to each mixture component. 'NbIterations' ; The number of performed iterations.

km20/gemalgorithm documentation built on May 29, 2019, 2:50 p.m.