aima.core.probability.hmm

Interface HiddenMarkovModel

All Known Implementing Classes:

HMM

public interface HiddenMarkovModel

Artificial Intelligence A Modern Approach (3rd Edition): page 578.

The hidden Markov model, or HMM. An HMM is a temporal probabilistic model in which the state of the process is described by a single discrete random variable. The possible values of the variable are the possible states of the world.

Author:

Ciaran O'Reilly, Ravi Mohan

Method Summary

Modifier and Type	Method and Description
Matrix	convert(CategoricalDistribution fromCD) Convert a Categorical Distribution into a column vector in Matrix form.
java.util.List<CategoricalDistribution>	convert(java.util.List<Matrix> matrixs) Convert a list of column vectors in Matrix form into a corresponding list of Categorical Distributions.
CategoricalDistribution	convert(Matrix fromMessage) Convert a column vector in Matrix form to a Categorical Distribution.
Matrix	createUnitMessage() Return a new column vector in matrix form with all values set to 1.0.
Matrix	getEvidence(java.util.List<AssignmentProposition> evidence) Get the specific evidence matrix based on assigned evidence value.
Matrix	getPrior() Return the prior distribution represented as a column vector in Matrix form.
java.util.Map<java.lang.Object,Matrix>	getSensorModel() Return the sensor model in matrix form: P(et | Xt = i) for each state i. For mathematical convenience we place each of these values into an S * S diagonal matrix.
RandomVariable	getStateVariable() Return the single discrete random variable used to describe the process state.
Matrix	getTransitionModel() Return the transition model: P(Xt | Xt-1) is represented by an S * S matrix T where Tij = P(Xt = j | Xt-1 = i).
Matrix	normalize(Matrix m) Create a normalized column vector in matrix form of the passed in column vector.

Method Detail

getStateVariable

RandomVariable getStateVariable()

Return the single discrete random variable used to describe the process state.

Returns:

the single discrete random variable used to describe the process state.

getTransitionModel

Matrix getTransitionModel()

Return the transition model:
P(X_t | X_t-1)
is represented by an S * S matrix T where
T_ij = P(X_t = j | X_t-1 = i).

Returns:

the transition model in Matrix form.

getSensorModel

java.util.Map<java.lang.Object,Matrix> getSensorModel()

Return the sensor model in matrix form:
P(e_t | X_t = i) for each state i.
For mathematical convenience we place each of these values into an S * S diagonal matrix.

Returns:

the sensor model in matrix form.

getPrior

Matrix getPrior()

Return the prior distribution represented as a column vector in Matrix form.

Returns:

the prior distribution represented as a column vector in Matrix form.

getEvidence

Matrix getEvidence(java.util.List<AssignmentProposition> evidence)

Get the specific evidence matrix based on assigned evidence value.

Parameters:

evidence - the evidence assignment e.

Returns:

the Matrix representation of this evidence assignment from the sensor model.

createUnitMessage

Matrix createUnitMessage()

Return a new column vector in matrix form with all values set to 1.0.

Returns:

a new column vector in matrix form with all values set to 1.0.

convert

Matrix convert(CategoricalDistribution fromCD)

Convert a Categorical Distribution into a column vector in Matrix form.

Parameters:

fromCD - the categorical distribution to be converted.

Returns:

a column vector in Matrix form of the passed in categorical distribution.

convert

CategoricalDistribution convert(Matrix fromMessage)

Convert a column vector in Matrix form to a Categorical Distribution.

Parameters:

fromMessage - the column vector in Matrix form to be converted.

Returns:

a categorical distribution representation of the passed in column vector.

convert

java.util.List<CategoricalDistribution> convert(java.util.List<Matrix> matrixs)

Convert a list of column vectors in Matrix form into a corresponding list of Categorical Distributions.

Parameters:

matrixs - the column vectors in matrix form to be converted.

Returns:

a corresponding list of Categorical Distribution representation of the passed in column vectors.

normalize

Matrix normalize(Matrix m)

Create a normalized column vector in matrix form of the passed in column vector.

Parameters:

m - a column vector representation in matrix form.

Returns:

a normalized column vector of the passed in column vector.