statistics - HMM's with independent training sets, Matlab -
i trying use hidden markov model (hmm) problem have m states , several independent training sets in matlab. each observation in each training set can allocated state, states not hidden such.
my question how can concatenate/join/use each independent training set train model , estimate transition probabilities of m states? preferable using matlab.
thanks
not sure if understood question or not. there no assumption training data in hmm. hmm training procedure considers every training sequence independent respect others.
so, guess if want represent different training sets unique hmm, have consider data set unique training set. way data has formatted depends on function/toolbox using training.
if using matlab, should have toolbox developed kevin murphy (ptmk toolbox) publicly available , easy use.
however, if know states of each data sample, , if looking @ transition matrix, should able compute hand. probability computation. (remember hmm works assumption state @ time (t+1) depends on state @ time t, , totally independent of previous states.)
Comments
Post a Comment