Fixed bolding of vectors x, x_t and x_n throughout the slides when used as a sample from multivariate Gaussian distribution or GMM.  Where a non-bolded x, x_t or x_n appears, it should be treated as a scalar.

On slides 26 and 29, an equation for variance was given, but it should have been for the covariance instead.

On slide 27, amended the EM equation to show that the log likelihood should be joint, p(x,m;lambda), rather than conditional.

On slide 38, one x was missing subscript t in the equation for covariance matrix estimation.