Class CharFromProbabilitiesOp

Provides outgoing messages for Char(Vector), given random arguments to the function.

Inheritance

Object

CharFromProbabilitiesOp

Inherited Members

Object.Equals(Object)

Object.Equals(Object, Object)

Object.GetHashCode()

Object.GetType()

Object.MemberwiseClone()

Object.ReferenceEquals(Object, Object)

Object.ToString()

Namespace: Microsoft.ML.Probabilistic.Factors

Assembly: Microsoft.ML.Probabilistic.dll

Syntax

[FactorMethod(typeof(Factor), "Char", new Type[]{})]
[Quality(QualityBand.Experimental)]
public static class CharFromProbabilitiesOp

Methods

CharacterAverageConditional(Dirichlet)

EP message to character.

Declaration

public static DiscreteChar CharacterAverageConditional(Dirichlet probabilities)

Parameters

Type	Name	Description
Dirichlet	probabilities	Incoming message from `probabilities`. Must be a proper distribution. If any element is uniform, the result will be uniform.

Returns

Type	Description
DiscreteChar	The outgoing EP message to the `character` argument.

Remarks

The outgoing message is a distribution matching the moments of character as the random arguments are varied. The formula is proj[p(character) sum_(probabilities) p(probabilities) factor(character,probabilities)]/p(character).

Exceptions

Type	Condition
ImproperMessageException	`probabilities` is not a proper distribution.

LogEvidenceRatio(Dirichlet, DiscreteChar)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(Dirichlet probabilities, DiscreteChar character)

Parameters

Type	Name	Description
Dirichlet	probabilities	Incoming message from `probabilities`.
DiscreteChar	character	Incoming message from `character`.

Returns

Type	Description
Double	Logarithm of the factor's contribution the EP model evidence.

Remarks

The formula for the result is log(sum_(probabilities,character) p(probabilities,character) factor(character,probabilities) / sum_character p(character) messageTo(character)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

LogEvidenceRatio(Dirichlet, Char)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(Dirichlet probabilities, char character)

Parameters

Type	Name	Description
Dirichlet	probabilities	Incoming message from `probabilities`.
Char	character	Constant value for `character`.

Returns

Type	Description
Double	Logarithm of the factor's contribution the EP model evidence.

Remarks

The formula for the result is log(sum_(probabilities) p(probabilities) factor(character,probabilities)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

ProbabilitiesAverageConditional(DiscreteChar, Dirichlet, Dirichlet)

EP message to probabilities.

Declaration

public static Dirichlet ProbabilitiesAverageConditional(DiscreteChar character, Dirichlet probabilities, Dirichlet result)

Parameters

Type	Name	Description
DiscreteChar	character	Incoming message from `character`. Must be a proper distribution. If uniform, the result will be uniform.
Dirichlet	probabilities	Incoming message from `probabilities`.
Dirichlet	result	Modified to contain the outgoing message.

Returns

Type	Description
Dirichlet	`result`

Remarks

The outgoing message is a distribution matching the moments of probabilities as the random arguments are varied. The formula is proj[p(probabilities) sum_(character) p(character) factor(character,probabilities)]/p(probabilities).

Exceptions

Type	Condition
ImproperMessageException	`character` is not a proper distribution.