EM.2: Expectation-maximization algorithm

[@gonzalopolo2612]

If you knew the labels/categories of the data (supervised learning), would this be like Linear Discriminant Analysis (same covariance matrix for each class) or Quadratic Discriminant Analysis (different Covariance Matrix)? Thank you very much, excellent videos as always.

[@dfrumkin]

Your variance estimator is biased and should have had n_b-1 in the denominator, right? (Of course, for a large number of samples, the bias is tiny).

[@tianyubai9230]

n in denominator is MLE and n-1 in the denominator is unbiased estimate. Here we want the MLE