Learning the parameters of an HMM
		Its easy to learn the parameters if , for each observed sequence of symbols, we can infer the posterior distribution across the sequences of hidden states    We can infer which hidden state sequence gave rise to an observed sequence by using the dynamic programming trick.