In the E-M algorithm, what is the purpose of the M-step?
To estimate the missing data.
To maximize the expected log-likelihood found in the E-step.
To calculate the posterior distribution of the parameters.
To determine the degrees of freedom for the next iteration.