[Topic-models] Regarding CTM-C

Mohammad Moghimi m.moghimi at gmail.com
Tue Nov 18 02:38:08 EST 2008


Dear All,


how can I compute P(d|z) for train and test documents in CTM model?
z=topic
d=document

After I have ran ctm with est and inf parameters the results dir contents
are as follows:
000-cov.dat
000-inv-cov.dat
000-lambda.dat
000-log-beta.dat
000-log-det-inv-cov.dat
000-mu.dat
000-nu.dat
000-param.txt
000.beta
000.other
005.beta
005.gamma
005.other
010-cov.dat
010-inv-cov.dat
010-lambda.dat
010-log-beta.dat
010-log-det-inv-cov.dat
010-mu.dat
010-nu.dat
010-param.txt
010.beta
010.gamma
010.other
015.beta
015.gamma
015.other
020.beta
020.gamma
020.other
025.beta
025.gamma
025.other
030.beta
030.gamma
030.other
final-cov.dat
final-inv-cov.dat
final-lambda.dat
final-log-beta.dat
final-log-det-inv-cov.dat
final-mu.dat
final-nu.dat
final-param.txt
final.beta
final.gamma
final.other
likelihood.dat
result-ctm-lhood.dat
result-lambda.dat
result-nu.dat
result-phi-sum.dat
result-word-assgn.dat
word-assignments.dat



Regards,
Mohammad Moghimi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/topic-models/attachments/20081118/58d4bbea/attachment.html>


More information about the Topic-models mailing list