|
|
Viterbi Algorithm1 2 3 4 5 The only difference is that the summation (  ) in the forward
algorithm is replaced with max to calculate the
 's - this
important difference picks out the most likely route to the
current position, rather than the total probability. We also,
for the Viterbi algorithm remember the best route to the current
position by maintaining a `back-pointer', via the argmax
calculation of the  's.
|