Mercurial > hg > camir-aes2014
annotate toolboxes/FullBNT-1.0.7/HMM/dhmm_logprob.m @ 0:e9a9cd732c1e tip
first hg version after svn
| author | wolffd |
|---|---|
| date | Tue, 10 Feb 2015 15:05:51 +0000 |
| parents | |
| children |
| rev | line source |
|---|---|
| wolffd@0 | 1 function [loglik, errors] = dhmm_logprob(data, prior, transmat, obsmat) |
| wolffd@0 | 2 % LOG_LIK_DHMM Compute the log-likelihood of a dataset using a discrete HMM |
| wolffd@0 | 3 % [loglik, errors] = log_lik_dhmm(data, prior, transmat, obsmat) |
| wolffd@0 | 4 % |
| wolffd@0 | 5 % data{m} or data(m,:) is the m'th sequence |
| wolffd@0 | 6 % errors is a list of the cases which received a loglik of -infinity |
| wolffd@0 | 7 |
| wolffd@0 | 8 if ~iscell(data) |
| wolffd@0 | 9 data = num2cell(data, 2); |
| wolffd@0 | 10 end |
| wolffd@0 | 11 ncases = length(data); |
| wolffd@0 | 12 |
| wolffd@0 | 13 loglik = 0; |
| wolffd@0 | 14 errors = []; |
| wolffd@0 | 15 for m=1:ncases |
| wolffd@0 | 16 obslik = multinomial_prob(data{m}, obsmat); |
| wolffd@0 | 17 [alpha, beta, gamma, ll] = fwdback(prior, transmat, obslik, 'fwd_only', 1); |
| wolffd@0 | 18 if ll==-inf |
| wolffd@0 | 19 errors = [errors m]; |
| wolffd@0 | 20 end |
| wolffd@0 | 21 loglik = loglik + ll; |
| wolffd@0 | 22 end |