search

jeffheaton / encog-java-core

http://www.heatonresearch.com/encog
Other
742 stars 268 forks source link

HMM mu & sigma #244

Open nirzohar opened 7 years ago

nirzohar commented 7 years ago

I use TrainBaumWelchScaled to train HiddenMarkovModel. It seems that the mu (mean) on the StateDistribution on the HiddenMarkovModel do not change after the iterations. Here is my logic within the for loop:

clac = (ForwardBackwardScaledCalculator) bwl.generateForwardBackwardCalculator(trainingSet, trainHmm); double curLoglik = clac.lnProbability();

bwl.iteration(); trainHmm = (HiddenMarkovModel) bwl.getMethod();

appreciate the help, or mail to nir@provenanceim.com

Thanks