mitkGradientDescentOptimizer Class Reference

mitkGradientDescentOptimizer - a concrete class for implementation of the Gradient Descent algorithm. More...

#include <mitkGradientDescentOptimizer.h>

Inherits mitkOptimizer.

Inheritance diagram for mitkGradientDescentOptimizer:

[legend]Collaboration diagram for mitkGradientDescentOptimizer:

[legend]List of all members.

Public Member Functions
virtual void	PrintSelf (ostream &os)
	mitkGradientDescentOptimizer ()
double	GetCurrentStepLength ()
double	GetMaximumStepLength ()
double	GetMinimumStepLength ()
void	SetCurrentStepLength (double length)
void	SetMaximumStepLength (double length)
void	SetMinimumStepLength (double length)
void	SetGradientMagnitudeTolerance (double tolerance)
void	SetRelaxationFactor (double factor)

---

Detailed Description

mitkGradientDescentOptimizer - a concrete class for implementation of the Gradient Descent algorithm.

mitkGradientDescentOptimizer - a concrete class for implementation of the Gradient Descent algorithm. Gradient Descent algorithm is a derivative based optimize method which needs image's gradient intensities to determine the optimization directions. In each recursive step, the algorithm will get current position in parameter space by evaluating the cost function and compute the next position following the equation:

nextPosition = currentPostion + direction * step length

The Gradient Descent algorithm is a simple, robust method and the convergence speed is fast.

---

Constructor & Destructor Documentation

mitkGradientDescentOptimizer::mitkGradientDescentOptimizer (   )

Constructor.

---

Member Function Documentation

double mitkGradientDescentOptimizer::GetCurrentStepLength (   )  [inline]

Get the current step length of optimization. Returns: Return the current step length.

double mitkGradientDescentOptimizer::GetMaximumStepLength (   )  [inline]

Get the maximum step length of optimization. Returns: Return the maximum step length.

double mitkGradientDescentOptimizer::GetMinimumStepLength (   )  [inline]

Get the minimum step length of optimization, implement in child class. Returns: Return the minimum step length.

virtual void mitkGradientDescentOptimizer::PrintSelf (  ostream &  os  )  [virtual]

Print the necessary information about this object for the debugging purpose. Parameters: os  The specified ostream to output information. Reimplemented from mitkOptimizer.

void mitkGradientDescentOptimizer::SetCurrentStepLength (  double  length  )  [inline]

Set the current step length of optimization. Parameters: length  The current step length.

void mitkGradientDescentOptimizer::SetGradientMagnitudeTolerance (  double  tolerance  )  [inline]

Set the gradient magnitude tolerance value. Parameters: tolerance  The tolerance value.

void mitkGradientDescentOptimizer::SetMaximumStepLength (  double  length  )  [inline]

Set the maximum step length of optimization. Parameters: length  The maximum step length.

void mitkGradientDescentOptimizer::SetMinimumStepLength (  double  length  )  [inline]

Set the minimum step length of optimization. Parameters: length  The minimum step length.

void mitkGradientDescentOptimizer::SetRelaxationFactor (  double  factor  )  [inline]

Set the relaxation factor for step updating. Parameters: factor  The factor value.

---

The documentation for this class was generated from the following file:

• mitkGradientDescentOptimizer.h

---