gov.sandia.cognition.learning.algorithm.gradient

Interface GradientDescendable

All Superinterfaces:

java.lang.Cloneable, CloneableSerializable, Evaluator<Vector,Vector>, ParameterGradientEvaluator<Vector,Vector,Matrix>, java.io.Serializable, VectorFunction, Vectorizable, VectorizableVectorFunction

All Known Implementing Classes:

DifferentiableFeedforwardNeuralNetwork, DifferentiableGeneralizedLinearModel, GradientDescendableApproximator, MultivariateDiscriminant, MultivariateDiscriminantWithBias, ThreeLayerFeedforwardNeuralNetwork

@CodeReview(reviewer="Kevin R. Dixon",date="2008-07-23",changesNeeded=false,comments={"Minor change to class javadoc.","Moved previous code review as CodeReview annotation","Looks fine."}) @CodeReview(reviewer="Justin Basilico",date="2006-10-04",changesNeeded=false,comments="Interface looks fine.")
public interface GradientDescendable
extends VectorizableVectorFunction, ParameterGradientEvaluator<Vector,Vector,Matrix>

Defines the functionality of an object that is required in order to apply the gradient descent algorithm to it. That is, an object that can differentiate its output (a vector) for a given input (another vector) with respect to its parameters (a matrix).

Since:

1.0

Author:

Justin Basilico, Kevin R. Dixon

Method Summary

Modifier and Type	Method and Description
GradientDescendable	clone() Creates a new clone (shallow copy) of this object.
Matrix	computeParameterGradient(Vector input) Computes the derivative of the function about the input with respect to the parameters of the function.

Methods inherited from interface gov.sandia.cognition.evaluator.Evaluator

evaluate

Methods inherited from interface gov.sandia.cognition.math.matrix.Vectorizable

convertFromVector, convertToVector

Method Detail

computeParameterGradient

Matrix computeParameterGradient(Vector input)

Computes the derivative of the function about the input with respect to the parameters of the function.

Specified by:

computeParameterGradient in interface ParameterGradientEvaluator<Vector,Vector,Matrix>

Parameters:

input - Point about which to differentiate w.r.t. the parameters.

Returns:

Matrix of parameter gradients.

clone

GradientDescendable clone()

Creates a new clone (shallow copy) of this object.

Specified by:

clone in interface CloneableSerializable

Specified by:

clone in interface Vectorizable

Specified by:

clone in interface VectorizableVectorFunction

Returns:

A new clone (shallow copy) of this object.