viterbi: Calculate Most Likely State Sequence Using the Viterbi...

In tileHMM: Hidden Markov Models for ChIP-on-Chip Analysis

Description

The Viterbi algorithm computes the most likely sequence of states given an HMM and an observation sequence.

Usage

## S4 method for signature 'hmm'
viterbi(hmm, obs, names=TRUE)

Arguments

hmm	Object of class hmm.
obs	A vector containing the observation sequence.
names	Logical indicating whether state names should be returned. If TRUE (the default) the returned sequence consists of state names, otherwise the state index is returned instead.

Value

A list with components

stateSeq	Most likely state sequence.
logProb	The probability of stateSeq given hmm and obs.
matrix	The dynamic programming matrix.

Author(s)

Peter Humburg

References

Viterbi, A. J. 1967 Error bounds for convolutional codes and an assymptotically optimal decoding algorithm. IEEE Transactions on Information Theory, 13, 2600–269.

See Also

viterbiTraining

Examples

## create two state HMM with t distributions
state.names <- c("one","two")
transition <- c(0.1, 0.02)
location <- c(1, 2)
scale <- c(1, 1)
df <- c(4, 6)
model <- getHMM(list(a=transition, mu=location, sigma=scale, nu=df),
    state.names)

## obtain observation sequence from model
obs <- sampleSeq(model, 100, return.states=TRUE)

## compute most likely state sequence for obs
vit.res <- viterbi(model, obs$observation)

## how well did we do?
sum(vit.res$stateSeq == obs$states)/length(vit.res$stateSeq)

Example output

[1] 0.85

tileHMM documentation built on May 30, 2017, 3:41 a.m.