hmm-fwbk {stochmod}R Documentation

Hidden Markov Models

Description

Forward-Backward procedure

Usage

HMM.fwbk( x, hmm )

Arguments

x [N x p] matrix of N samples in p dimensions
hmm An HMM

Details

The E-step of the EM algorithm for Hidden Markov Models

Value

alpha [N x K] matrix of forward variable values
beta [N x K] matrix of backward variable values
c [N x 1] vector of scaling coefficients
gamma [N x K] matrix of transition probabilities from state i: gamma[t,i] = P( Qt = i | O,hmmp )
xi [N-1 x K x K] array of transition probabilties from state i to state j: xi[t,i,j] = P( Qt = i, Qt+1 = j | O,hmmp )

Author(s)

Artem Sokolov Artem.Sokolov@gmail.com


[Package stochmod version 1.2 Index]