#include <itkGoodnessOfFitMixtureModelCostFunction.h>
Inheritance diagram for itk::Statistics::GoodnessOfFitMixtureModelCostFunction:
Public Types | |
typedef GoodnessOfFitMixtureModelCostFunction | Self |
typedef SingleValuedCostFunction | Superclass |
typedef SmartPointer< Self > | Pointer |
typedef SmartPointer< const Self > | ConstPointer |
typedef TInputSample | InputSampleType |
typedef TInputSample::MeasurementType | MeasurementType |
typedef TInputSample::MeasurementVectorType | MeasurementVectorType |
typedef SingleValuedCostFunction::ParametersType | ParamtersType |
typedef SingleValuedCostFunction::MeasureType | MeasureType |
typedef GoodnessOfFitComponentBase< TInputSample > | ComponentType |
typedef std::vector< ComponentType * > | ComponentVectorType |
typedef GoodnessOfFitFunctionBase< typename ComponentType::HistogramType > | FunctionType |
Public Methods | |
virtual const char * | GetClassName () const |
itkStaticConstMacro (MeasurementVectorSize, unsigned int, TInputSample::MeasurementVectorSize) | |
void | AddComponent (ComponentType *component) |
void | SetFunction (FunctionType *core) |
FunctionType * | GetFunction () |
virtual unsigned int | GetNumberOfParameters () const |
virtual MeasureType | GetValue (const ParametersType ¶meters) const |
virtual void | GetDerivative (const ParametersType ¶meters, DerivativeType &derivative) const |
Static Public Methods | |
Pointer | New () |
Protected Methods | |
GoodnessOfFitMixtureModelCostFunction () | |
virtual | ~GoodnessOfFitMixtureModelCostFunction () |
virtual void | PrintSelf (std::ostream &os, Indent indent) const |
The goodness-of-fit statistics for a single model is discrepancy between the observed frequency and the expected frequency. To reduce computational load of multivariate case, this class uses projective method.
The projective multivariate goodness-of-fit statistics calculation follows the following steps:
1) creates a subsample that includes the measurement vectors that fall in a spherical kernel. 2) finds the base axes determined by the eigen vectors of the covariance matrix. 3) project the subsample on to one of the base axes (from step 2) 4) calculates the observed frequencies (in an 1D Histogram object) after projection (step 3) and the expected frequencies (in an 1D Histogram object) 5) calculates the discrepancy between the observed histogram and the expected histogram using a goodness-of-fit statistics 6) repeat step 3) - 5) and sum the goodness-of-fit values
For a mixture model, the above procedure is applied independently for each model (module). The sum of the goodness-of-fit values of models is the goodness-of-fit statistics for the mixture model.
The step 1) - 4) is done by the subclasses of GoodnessOfFitComponentBase, and the step 5) is done by the subclasses of GoodnessOfFitFunctionBase.
To see how this class interacts GoodnessOfFitComponentBase objects and GoodnessOfFitFunctionBase objects, please look at the implementation of the GetValue method of this class.
Better fit means smaller goodness-of-fit value in this implementation. This class is following the SingleValuedCostFunction interfaces so that users can uses this function with any subclasses of SingleValuedNonLinearOptimizer class as long as they do not use GetDerivative and GetValueAndDerivative methods.
Definition at line 75 of file itkGoodnessOfFitMixtureModelCostFunction.h.
|
Definition at line 105 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Definition at line 106 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Reimplemented from itk::SingleValuedCostFunction. Definition at line 83 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Definition at line 109 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Definition at line 91 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Definition at line 94 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Definition at line 95 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
MeasureType typedef. It defines a type used to return the cost function value. Reimplemented from itk::SingleValuedCostFunction. Definition at line 103 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
ParametersType typedef. It defines a position in the optimization search space. Definition at line 99 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Reimplemented from itk::SingleValuedCostFunction. Definition at line 82 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Standard class typedefs Reimplemented from itk::SingleValuedCostFunction. Definition at line 80 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Reimplemented from itk::SingleValuedCostFunction. Definition at line 81 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
|
|
|
|
aceesing methods for the sample manipulator |
|
Run-time type information (and related methods). Reimplemented from itk::SingleValuedCostFunction. |
|
This method returns the derivative of the cost function corresponding to the specified parameters. Implements itk::SingleValuedCostFunction. Definition at line 128 of file itkGoodnessOfFitMixtureModelCostFunction.h. References itk::DerivativeType, and itk::ParametersType. |
|
Definition at line 117 of file itkGoodnessOfFitMixtureModelCostFunction.h. |
|
Return the number of parameters required to compute this cost function. This method MUST be overloaded by derived classes. Implements itk::CostFunction. |
|
This method returns the value of the cost function corresponding to the specified parameters. Implements itk::SingleValuedCostFunction. |
|
|
|
Method for creation through the object factory. Reimplemented from itk::Object. |
|
Methods invoked by Print() to print information about the object including superclasses. Typically not called by the user (use Print() instead) but used in the hierarchical print process to combine the output of several classes. Reimplemented from itk::CostFunction. |
|
aceesing methods for the expected probability histogram |