1) Just do EM with parameters in an arbitrary semiring and you generalize nicely to all the special cases you mentioned (and more) 2) I see how all of EM is co

1) Just do EM with parameters in an arbitrary semiring and you generalize nicely to all the special cases you mentioned (and more) 2) I see how all of EM is coordinate ascent, but I don't see how the M step is gradient ascent. Where the gradients at? 3) Second-order optimization methods can be pretty tricky and fall apart for large problems. Stochastic gradient descent ftw.