Logo ROOT  
Reference Guide
 
Loading...
Searching...
No Matches
TMVA::ConvergenceTest Class Reference

Check for convergence.

Definition at line 46 of file ConvergenceTest.h.

Public Member Functions

 ConvergenceTest ()
 constructor
 
 ~ConvergenceTest ()
 destructor
 
Float_t GetCurrentValue ()
 
Bool_t HasConverged (Bool_t withinConvergenceBand=kFALSE)
 gives back true if the last "steps" steps have lead to an improvement of the "fitness" of the "individuals" of at least "improvement"
 
Float_t Progress ()
 returns a float from 0 (just started) to 1 (finished)
 
void ResetConvergenceCounter ()
 
void SetConvergenceParameters (Int_t steps, Double_t improvement)
 
void SetCurrentValue (Float_t value)
 
Float_t SpeedControl (UInt_t ofSteps)
 this function provides the ability to change the learning rate according to the success of the last generations.
 

Protected Attributes

Float_t fCurrentValue
 ! current value
 
Float_t fImprovement
 ! minimum improvement which counts as improvement
 
Int_t fSteps
 ! number of steps without improvement required for convergence
 

Private Attributes

Float_t fBestResult
 
Float_t fConvValue
 ! the best "fitness" value
 
Int_t fCounter
 ! counts the number of steps without improvement
 
Float_t fLastResult
 
Int_t fMaxCounter
 ! maximum value for the counter so far
 
std::deque< Short_tfSuccessList
 to calculate the improvement-speed
 

#include <TMVA/ConvergenceTest.h>

Inheritance diagram for TMVA::ConvergenceTest:
[legend]

Constructor & Destructor Documentation

◆ ConvergenceTest()

TMVA::ConvergenceTest::ConvergenceTest ( )

constructor

Definition at line 40 of file ConvergenceTest.cxx.

◆ ~ConvergenceTest()

TMVA::ConvergenceTest::~ConvergenceTest ( )

destructor

Definition at line 55 of file ConvergenceTest.cxx.

Member Function Documentation

◆ GetCurrentValue()

Float_t TMVA::ConvergenceTest::GetCurrentValue ( )
inline

Definition at line 57 of file ConvergenceTest.h.

◆ HasConverged()

Bool_t TMVA::ConvergenceTest::HasConverged ( Bool_t  withinConvergenceBand = kFALSE)

gives back true if the last "steps" steps have lead to an improvement of the "fitness" of the "individuals" of at least "improvement"

this gives a simple measure of if the estimator of the MLP is converging and no major improvement is to be expected.

Definition at line 66 of file ConvergenceTest.cxx.

◆ Progress()

Float_t TMVA::ConvergenceTest::Progress ( )

returns a float from 0 (just started) to 1 (finished)

Definition at line 91 of file ConvergenceTest.cxx.

◆ ResetConvergenceCounter()

void TMVA::ConvergenceTest::ResetConvergenceCounter ( )
inline

Definition at line 58 of file ConvergenceTest.h.

◆ SetConvergenceParameters()

void TMVA::ConvergenceTest::SetConvergenceParameters ( Int_t  steps,
Double_t  improvement 
)
inline

Definition at line 54 of file ConvergenceTest.h.

◆ SetCurrentValue()

void TMVA::ConvergenceTest::SetCurrentValue ( Float_t  value)
inline

Definition at line 56 of file ConvergenceTest.h.

◆ SpeedControl()

Float_t TMVA::ConvergenceTest::SpeedControl ( UInt_t  ofSteps)

this function provides the ability to change the learning rate according to the success of the last generations.

Parameters:

  • int ofSteps : = if OF the number of STEPS given in this variable (ofSteps) the rate of improvement has to be calculated

using this function one can increase the stepSize of the mutation when we have good success (to pass fast through the easy phase-space) and reduce the learning rate if we are in a difficult "territory" of the phase-space.

Definition at line 112 of file ConvergenceTest.cxx.

Member Data Documentation

◆ fBestResult

Float_t TMVA::ConvergenceTest::fBestResult
private

Definition at line 81 of file ConvergenceTest.h.

◆ fConvValue

Float_t TMVA::ConvergenceTest::fConvValue
private

! the best "fitness" value

Definition at line 76 of file ConvergenceTest.h.

◆ fCounter

Int_t TMVA::ConvergenceTest::fCounter
private

! counts the number of steps without improvement

Definition at line 75 of file ConvergenceTest.h.

◆ fCurrentValue

Float_t TMVA::ConvergenceTest::fCurrentValue
protected

! current value

Definition at line 68 of file ConvergenceTest.h.

◆ fImprovement

Float_t TMVA::ConvergenceTest::fImprovement
protected

! minimum improvement which counts as improvement

Definition at line 70 of file ConvergenceTest.h.

◆ fLastResult

Float_t TMVA::ConvergenceTest::fLastResult
private

Definition at line 82 of file ConvergenceTest.h.

◆ fMaxCounter

Int_t TMVA::ConvergenceTest::fMaxCounter
private

! maximum value for the counter so far

Definition at line 77 of file ConvergenceTest.h.

◆ fSteps

Int_t TMVA::ConvergenceTest::fSteps
protected

! number of steps without improvement required for convergence

Definition at line 71 of file ConvergenceTest.h.

◆ fSuccessList

std::deque<Short_t> TMVA::ConvergenceTest::fSuccessList
private

to calculate the improvement-speed

Definition at line 83 of file ConvergenceTest.h.

Libraries for TMVA::ConvergenceTest:

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