Next: 捏捌缄恕の网爬 Up: High order Joint Probability(捏捌缄恕) Previous: High order Joint Probability(捏捌缄恕) 鐩

捏捌缄恕の车妥

塑甫垫では·咐胳モデルに

-gramモデルの洛わりにHigh order Joint Probability∈笆布Joint Probability∷を脱いるˉ笆布の及(2)にその纷换及を绩す[2]ˉ

$\begin{eqnarray*} \hspace{-3cm}\sum_{j=0}^{M-1}\sum_{i=0}^{N-1}P(J_{j-2},J_{j-1},J_j,E_{i-2},E_{i-1},E_i) \end{eqnarray*}$

$\begin{eqnarray*}\hspace{-1cm}\times\log_2\frac{P(J_{j-2},J_{j-1},J_j,E_{i-2},E_{i-1},E_i)}{P(J_{j-2},J_{j-1},J_j)P(E_{i-2},E_{i-1},E_i)} \end{eqnarray*}$

$\begin{eqnarray*} \hspace{-2cm} =\sum_{j=0}^{M-1}\sum_{i=0}^{N-1}\frac{count(J_{j-2},J_{j-1},J_j,E_{i-2},E_{i-1},E_i)}{N_{je}} \end{eqnarray*}$

$\displaystyle \hspace{-1.4cm}\times\log_2{\frac{\frac{count(J_{j-2},J_{j-1},J_j... ...{\frac{count(J_{j-2},J_{j-1},J_j)}{N_j}\frac{count(E_{i-2},E_{i-1},E_i)}{N_e}}}$