S sina786 New Member Jan 2, 2013 #1 Hi I have a question about statistical machine translation. in SMT, we use Bayesian rule to compute p(E|F) as P(F|E)*P(E). why we don't directly calculate p(E|F) as the same way that we calculate p(F|E)? why we convert p(E|F) to P(F|E)*P(E)? Thanks
Hi I have a question about statistical machine translation. in SMT, we use Bayesian rule to compute p(E|F) as P(F|E)*P(E). why we don't directly calculate p(E|F) as the same way that we calculate p(F|E)? why we convert p(E|F) to P(F|E)*P(E)? Thanks
