Implementation of a Hidden Markov model on a feature Vector as well as calculating distance between two HMMs

Hi,

Basically, a HMM is charecterized by the triplet (A,B,π).

I have read this paper * http://www.irisa.fr/vista/Papers/2007_icip_hervieu.pdf * and was trying to implement Hidden Markov model on my feature vector.

I have my Feature vector(row vector) as V = [V1 V2 V3 V4....Vn]. I want to do the following:

1) I want to distribute the values in the V,on an interval containing a given percentage P,within a domain [-S,S] (which is divided into N number of bins).

2) Let us assume that the no. of states are M.I want to calculate the A, π and B matrices using a least square technique.

3) Once,we calculate the parameters A, π and B for Feature vector 1 as well as Feature vector 2, I want to calculate the distance,inorder to compare two HMMs,proposed by Ranber.

I would be really glad,if someone can help me with the code. Thanks in advance.

[Athal Saleh]

Were you able to find a solution? because I've been facing the same issue too.