Package gov.sandia.cognition.learning.function.cost

Provides cost functions.

See:
          Description

Interface Summary
CostFunction<EvaluatedType,CostParametersType> The CostFunction interface defines the interface to evaluate some object to determine its cost.
DifferentiableCostFunction The DifferentiableCostFunction is a cost function that can be differentiated.
ParallelizableCostFunction Interface describing a cost function that can (largely) be computed in parallel.
SupervisedCostFunction<InputType,TargetType> A type of CostFunction normally used in supervised-learning applications.
 

Class Summary
AbstractCostFunction<EvaluatedType,CostParametersType> Partial implementation of CostFunction.
AbstractParallelizableCostFunction Partial implementation of the ParallelizableCostFunction
AbstractSupervisedCostFunction<InputType,TargetType> Partial implementation of SupervisedCostFunction
ClusterDistortionMeasure<DataType,ClusterType extends Cluster<DataType>> Computes the objective measure for a clustering algorithm, based on the internal "distortion" of each cluster.
EuclideanDistanceCostFunction The EuclideanDistanceCostFunction class implements a CostFunction that calculates the Euclidean distance the given Vectorizable and the goal vector.
KolmogorovSmirnovDivergence<DataType extends Number> CostFunction that induces a CDF that most-closely resembles the target distribution according to the Kolmogorov-Smirnov (K-S) test.
MeanL1CostFunction Cost function that evaluates the mean 1-norm error (absolute value of difference) weighted by a sample "weight" that is embedded in each sample.
MeanSquaredErrorCostFunction The MeanSquaredErrorCostFunction implements a cost function for functions that take as input a vector and return a vector.
NegativeLogLikelihood<DataType> CostFunction for computing the maximum likelihood (because we are minimizing the negative of the log likelihood)
ParallelClusterDistortionMeasure<DataType,ClusterType extends Cluster<DataType>> A parallel implementation of ClusterDistortionMeasure.
ParallelizedCostFunctionContainer A cost function that automatically splits a ParallelizableCostFunction across multiple cores/processors to speed up computation.
ParallelizedCostFunctionContainer.SubCostEvaluate Callable task for the evaluate() method.
ParallelizedCostFunctionContainer.SubCostGradient Callable task for the computeGradient() method
ParallelNegativeLogLikelihood<DataType> Parallel implementation of the NegativeLogLikleihood cost function
ParallelNegativeLogLikelihood.NegativeLogLikelihoodTask<DataType> Task for computing partial log likelihoods
SumSquaredErrorCostFunction This is the sum-squared error cost function
SumSquaredErrorCostFunction.Cache Caches often-used values for the Cost Function
SumSquaredErrorCostFunction.GradientPartialSSE Partial result from the SSE gradient computation
 

Package gov.sandia.cognition.learning.function.cost Description

Provides cost functions. Cost functions evaluate the cost of an object with resepect to some set of parameters, which are typically some form of data.

Since:
2.0
Author:
Justin Basilico