From: nasim on
i know for a HMM need to solve three problem

i have this parameter:
o=3; q=3;
init SMatrix(1*3) , trans Matrix(3*3), emis Matrix(3*3)
-----------------------
in the HMM toolBox of this site i have this code:
http://www.cs.ubc.ca/~murphyk/Software/HMM/hmm.html

1)learning
T=10;
nex=20;
data = dhmm_sample(init, trans, emis, nex, T);
[LL, prior, transmat, obsmat] = dhmm_em(data, init, trans, emis, 'max_iter', 5);
2)evaluate
loglik = dhmm_logprob(data, prior, transmat, obsmat)
3)mputing the most probable sequence (Viterbi)
B = multinomial_prob(data, obsmat);
[path] = viterbi_path(prior, transmat, B)

----------------------
1- learning:
seq1 = hmmgenerate(100,trans,emis);
[estTR,estE] = hmmtrain(seqs,trans,emis);

2-evaluate
[estimateTR,estimateE] = hmmestimate(seq,states);

3-decode
estimatedStates = hmmviterbi(seq,trans,emis);

is it correct?
how to set the init matrix like in first code?
is it possible guide me about the hmmdecode function, i don't understand the usage of it;