Hidden Markov Models

Definition
Usages
Summary
Section 1 - Page 1

Definition of a hidden Markov model

A hidden Markov model (HMM) is a triple (P,A,B).

P : the vector of the initial state probabilities;
A : the state transition matrix;
B : the confusion matrix;

Each probability in the state transition matrix and in the confusion matrix is time independent - that is, the matrices do not change in time as the system evolves. In practice, this is one of the most unrealistic assumptions of Markov models about real processes.