CODEWITHSUNDEEP
modelhiddenmarkov
Ali
Ali

Reputation: 1

Hidden Markov model for classification

I am using the hidden markov model for classification problems. I need to examine the impact of number of hidden states and number of visible states on the performance of classifier? Based on my results, the increase of number of hidden or visible states goes up the performance of model. Are these results meaningful? Why?

Upvotes: 0

Views: 229

Answers (1)

Magnus Berg Sletfjerding
Magnus Berg Sletfjerding

Reputation: 326

Depending on your data + initial parameters you provide (transition probability, maximum number of allowed states), the Hidden Markov Model will find a set number of states. If your data has a lot of incremental increases, it will most likely be better represented by a HMM with more states. If you provide some data / example it might be easier to help you.

Upvotes: 1

Related Questions