Question 2 Assume we wish to use the EM algorithm to maximize the posterior distribution over pa- rameters p(@Y) for a model containing latent variables Z and where Y is the observed data set. 1. Describe the modification of the E-step compared to the maximum likelihood case. 2. Show that the quantity to be maximized in the M-step is given by Q (0,0i-1) + log p (0) where = Q (0,01-1) = logp (Y,Z|0) p (Z|Y, 03-1) (- ( Question 2 Assume we wish to use the EM algorithm to maximize the posterior distribution over pa- rameters p(@Y) for a model containing latent variables Z and where Y is the observed data set. 1. Describe the modification of the E-step compared to the maximum likelihood case. 2. Show that the quantity to be maximized in the M-step is given by Q (0,0i-1) + log p (0) where = Q (0,01-1) = logp (Y,Z|0) p (Z|Y, 03-1) (- (