Introduction
Hidden Markov Models (HMMs) are widely used in science [1,2], engineering [3,5] and many other areas [4].
In a HMM, there are two types of states: the observable states and the hidden states.
For both types of the states, the underlying process is a Markov chain process [6].
To define a HMM, one has to define the number of both types of states
and also their transition probabilities. Very often, the transition probabilities of the
observable states depend on the hidden states. One major problem here is to determine
the transition probabilities of the hidden states because the transitions among the hidden
states are supposed to be unobservable. Here we propose a simple estimation method for
the transition probabilities among the hidden states.
|