HMM viterbi algorithm linked to values

Hi all,

I have used the HMM toolbox, using 4 states (Each states can take numbers from 1 to 6).

Then I apply the viterbi algortihm and find the optimum path of states. However, my question is:

How then the viterbi states of a given sequence are linked to certain values? here are the input data

A=[0.6 0.2 0.1 0.1;0.1 0.6 0.15 0.15; 0.15 0.15 0.6 0.1; 0.1 0.1 0.25 0.55]; %transition matrix

emis = [0.025 0.135 0.34 0.34 0.135 0.025;0.025 0.135 0.34 0.34 0.135 0.025;0.025 0.135 0.34 0.34 0.135 0.025;0.025 0.135 0.34 0.34 0.135 0.025]; %emission matrix, probabilities follow gaussian distribution (six bins of probabilities)

Initial_vales= [ 0.25 0.32 0.58 0.42 ..0.8]; %initial data given (24*20), range from 0.1 to 1.8!

%states_1= 0-0.25

%state_2=0.25-0.50

%state_2=0.5-0.75

%state_4=>0.75

thanks in advance,

Nikolas

Answer 1

[Aneela]

Hi Nikolas Spiliopoulos,

The Viterbi algorithm generates the most likely sequence of the hidden states (Viterbi States) in “HMM (Hidden Markov Models)”.

• The algorithm begins by determining the initial likelihoods for each state, based on the first observation.
• For each subsequent observation, the algorithm calculates the probability of being in each state by using the new observation and the previous state probabilities.

The Viterbi states of a given sequence are linked to the certain values through the emission probabilities of the observed value and the transition probabilities from the previous state.