Closed changediyasunny closed 7 years ago
s' is a single state. This definition is specifying how at time t we must consider all possible prior states when computing the max.
For the final state, we just compute the max over the final states in the sequence. Note that in this assignment, you do not need to add artificial "end states."
-Aron
On Sun, Feb 19, 2017 at 11:30 PM, schanged notifications@github.com wrote:
Hi Prof, I have couple of doubts implementing viterbi algorithm:
1.
What is the sequence of states S' in viterbi[s,t]←N max s'=1 {viterbi[s',t −1] ∗ a[s',s] ∗ bs(O[t])} 2.
How to calculate viterbi[qf, t] ? viterbi[qF ,T]←N max s'=1 viterbi[s,T] ∗ a[s,qF] How to take final state ? ` This algorithm is getting heavy and very confusing.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/iit-cs585/assignments/issues/15, or mute the thread https://github.com/notifications/unsubscribe-auth/ADv-hdNaFIwGththpgmDqmybmIMKVCQRks5reSTZgaJpZM4MFxyM .
alright. thank you Prof.
Hi Prof, I have couple of doubts implementing viterbi algorithm:
What is the sequence of states S' in
viterbi[s,t]←N max s'=1 {viterbi[s',t −1] ∗ a[s',s] ∗ bs(O[t])}
How to calculate viterbi[qf, t] ?
viterbi[qF ,T]←N max s'=1 viterbi[s,T] ∗ a[s,qF]
How to take final state ? ` This algorithm is getting heavy and very confusing.