GClasses::GOptimizer Class Reference

Detailed Description

This is the base class of all search algorithms that can jump to any vector in the search space seek the vector that minimizes error.

#include <GOptimizer.h>

Inheritance diagram for GClasses::GOptimizer:

Public Member Functions
	GOptimizer (GTargetFunction *pCritic)
virtual	~GOptimizer ()
void	basicTest (double minAccuracy, double warnRange=0.001)
	This is a helper method used by the unit tests of several model learners. More...
virtual const GVec &	currentVector ()=0
	Returns the current vector of the optimizer. For greedy search methods, this will be the best vector yet found. More...
virtual double	iterate ()=0
	Makes another attempt to find a better vector. Returns the heuristic error. (Usually you will call this method in a loop until your stopping criteria has been met.) More...
double	searchUntil (size_t nBurnInIterations, size_t nIterations, double dImprovement)
	This will first call iterate() nBurnInIterations times, then it will repeatedly call iterate() in blocks of nIterations times. If the error heuristic has not improved by the specified ratio after a block of iterations, it will stop. (For example, if the error before the block of iterations was 50, and the error after is 49, then training will stop if dImprovement is > 0.02.) If the error heuristic is not stable, then the value of nIterations should be large. More...

Protected Attributes
GTargetFunction *	m_pCritic

Constructor & Destructor Documentation

GClasses::GOptimizer::GOptimizer

(

GTargetFunction *

pCritic

)

virtual GClasses::GOptimizer::~GOptimizer ( )

virtual

Member Function Documentation

void GClasses::GOptimizer::basicTest	(	double	minAccuracy,
		double	warnRange = 0.001
	)

This is a helper method used by the unit tests of several model learners.

virtual const GVec& GClasses::GOptimizer::currentVector ( )

pure virtual

Returns the current vector of the optimizer. For greedy search methods, this will be the best vector yet found.

Implemented in GClasses::GSampleClimber, GClasses::GEmpiricalGradientDescent, GClasses::GRandomDirectionBinarySearch, GClasses::GAnnealing, GClasses::GProbeSearch, GClasses::GMinBinSearch, GClasses::GHillClimber, GClasses::GBouncyBalls, GClasses::GRandomSearch, GClasses::GMomentumGreedySearch, GClasses::GGridSearch, and GClasses::GEvolutionaryOptimizer.

virtual double GClasses::GOptimizer::iterate ( )

pure virtual

Makes another attempt to find a better vector. Returns the heuristic error. (Usually you will call this method in a loop until your stopping criteria has been met.)

Implemented in GClasses::GSampleClimber, GClasses::GEmpiricalGradientDescent, GClasses::GRandomDirectionBinarySearch, GClasses::GAnnealing, GClasses::GProbeSearch, GClasses::GHillClimber, GClasses::GMinBinSearch, GClasses::GBouncyBalls, GClasses::GRandomSearch, GClasses::GMomentumGreedySearch, GClasses::GEvolutionaryOptimizer, GClasses::GParticleSwarm, and GClasses::GGridSearch.

double GClasses::GOptimizer::searchUntil	(	size_t	nBurnInIterations,
		size_t	nIterations,
		double	dImprovement
	)

This will first call iterate() nBurnInIterations times, then it will repeatedly call iterate() in blocks of nIterations times. If the error heuristic has not improved by the specified ratio after a block of iterations, it will stop. (For example, if the error before the block of iterations was 50, and the error after is 49, then training will stop if dImprovement is > 0.02.) If the error heuristic is not stable, then the value of nIterations should be large.

Member Data Documentation

GTargetFunction* GClasses::GOptimizer::m_pCritic

protected