Class VectorSoftmaxOp_KM11

Provides outgoing messages for Softmax(IList<Double>), given random arguments to the function.

Inheritance

Object

VectorSoftmaxOp_KM11

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(MMath), "Softmax", new Type[]{typeof(Vector)})]
[Quality(QualityBand.Preview)]
[Buffers(new string[]{"A"})]
public static class VectorSoftmaxOp_KM11

Remarks

This implementation uses the bound in Knowles and Minka (2011), followed by nonconjugate VMP. This approach is linear in the dimension K.

Methods

A(VectorGaussian, Vector)

Update the buffer A.

Declaration

public static Vector A(VectorGaussian x, Vector a)

Parameters

Type	Name	Description
VectorGaussian	x	Incoming message from x. Must be a proper distribution. If any element is uniform, the result will be uniform.
Vector	a	Buffer a.

Returns

Type	Description
Vector	New value of buffer A.

Remarks

Exceptions

Type	Condition
ImproperMessageException	x is not a proper distribution.

AInit(VectorGaussian)

Initialize the buffer A.

Declaration

public static Vector AInit(VectorGaussian x)

Parameters

Type	Name	Description
VectorGaussian	x	Incoming message from x.

Returns

Type	Description
Vector	Initial value of buffer A.

Remarks

AverageLogFactor()

Evidence message for VMP.

Declaration

public static double AverageLogFactor()

Returns

Type	Description
Double	Zero.

Remarks

The formula for the result is log(factor(softmax,x)). Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

SoftmaxAverageLogarithm(VectorGaussian, Vector, Dirichlet)

VMP message to softmax.

Declaration

public static Dirichlet SoftmaxAverageLogarithm(VectorGaussian x, Vector a, Dirichlet result)

Parameters

Type	Name	Description
VectorGaussian	x	Incoming message from x. Must be a proper distribution. If any element is uniform, the result will be uniform.
Vector	a	Buffer a.
Dirichlet	result	Modified to contain the outgoing message.

Returns

Type	Description
Dirichlet	result

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	x is not a proper distribution.

SoftmaxAverageLogarithmInit(VectorGaussian)

Declaration

public static Dirichlet SoftmaxAverageLogarithmInit(VectorGaussian x)

Parameters

Type	Name	Description
VectorGaussian	x	Incoming message from x.

Returns

Type	Description
Dirichlet

Remarks

VectorGaussianToGaussianList(VectorGaussian)

Declaration

public static Gaussian[] VectorGaussianToGaussianList(VectorGaussian x)

Parameters

Type	Name	Description
VectorGaussian	x

Returns

Type	Description
Gaussian[]

XAverageLogarithm(Dirichlet, VectorGaussian, Vector, VectorGaussian)

VMP message to x.

Declaration

public static VectorGaussian XAverageLogarithm(Dirichlet softmax, VectorGaussian x, Vector a, VectorGaussian result)

Parameters

Type	Name	Description
Dirichlet	softmax	Incoming message from softmax. Must be a proper distribution. If any element is uniform, the result will be uniform.
VectorGaussian	x	Incoming message from x. Must be a proper distribution. If any element is uniform, the result will be uniform.
Vector	a	Buffer a.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of x with softmax integrated out. The formula is sum_softmax p(softmax) factor(softmax,x).

Exceptions

Type	Condition
ImproperMessageException	softmax is not a proper distribution.
ImproperMessageException	x is not a proper distribution.