IGLib  1.5
The IGLib base library for development of numerical, technical and business applications.
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Properties Events
IG.Num.NeuralTrainingParameters.ComparerBase Class Reference

Base comparer class (implementation of the IComparer<NeuralTrainingParameters> interface) for conmparing objects of type NeuralTrainingParameters More...

+ Inheritance diagram for IG.Num.NeuralTrainingParameters.ComparerBase:
+ Collaboration diagram for IG.Num.NeuralTrainingParameters.ComparerBase:

Protected Member Functions

double OutputNorm (IVector vec)
 

Protected Attributes

IVector _outputScalingLengths
 
bool _compareMinError = false
 
int _numLastErrors = 1
 
int _numBundles = 0
 
bool _compareByTrainingRmsError = true
 
bool _compareByTrainingMaxError = false
 
bool _compareByVerificationRmsError = false
 
bool _compareByVerificationMaxError = false
 

Properties

IVector OutputScalingLengths [get, set]
 Vector of scaling lengths for calculation of weighted norms. More...
 
bool CompareMinError [get, set]
 Whether the min error in convergence table is used for comparison when errors are compared. If false then mean value of last errors is used. More...
 
int NumLastErrors [get, set]
 Number of last errors in convergence list for calculataing the mean value of error. The default valeu is 1 which represent the last error in the converhence list. More...
 
int NumBundles [get, set]
 Number of bundles where sorting of convergences stars. The default valeu is 0 which represent normal sorting after the training is done. More...
 
bool CompareByTrainingRmsError [get, set]
 Whether training RMS errors from convergence tavble are compared. More...
 
bool CompareByTrainingMaxError [get, set]
 Whether Maximal absolute training errors from convergence tavble are compared. More...
 
bool CompareByVerificationRmsError [get, set]
 Whether verification RMS errors from convergence tavble are compared. More...
 
bool CompareByVerificationMaxError [get, set]
 Whether Maximal absolute verification errors from convergence tavble are compared. More...
 

Private Member Functions

int IComparer
< NeuralTrainingParameters >. 
Compare (NeuralTrainingParameters a, NeuralTrainingParameters b)
 

Detailed Description

Base comparer class (implementation of the IComparer<NeuralTrainingParameters> interface) for conmparing objects of type NeuralTrainingParameters

Member Function Documentation

double IG.Num.NeuralTrainingParameters.ComparerBase.OutputNorm ( IVector  vec)
inlineprotected
int IComparer<NeuralTrainingParameters>. IG.Num.NeuralTrainingParameters.ComparerBase.Compare ( NeuralTrainingParameters  a,
NeuralTrainingParameters  b 
)
inlineprivate

Member Data Documentation

IVector IG.Num.NeuralTrainingParameters.ComparerBase._outputScalingLengths
protected
bool IG.Num.NeuralTrainingParameters.ComparerBase._compareMinError = false
protected
int IG.Num.NeuralTrainingParameters.ComparerBase._numLastErrors = 1
protected
int IG.Num.NeuralTrainingParameters.ComparerBase._numBundles = 0
protected
bool IG.Num.NeuralTrainingParameters.ComparerBase._compareByTrainingRmsError = true
protected
bool IG.Num.NeuralTrainingParameters.ComparerBase._compareByTrainingMaxError = false
protected
bool IG.Num.NeuralTrainingParameters.ComparerBase._compareByVerificationRmsError = false
protected
bool IG.Num.NeuralTrainingParameters.ComparerBase._compareByVerificationMaxError = false
protected

Property Documentation

IVector IG.Num.NeuralTrainingParameters.ComparerBase.OutputScalingLengths
getset

Vector of scaling lengths for calculation of weighted norms.

bool IG.Num.NeuralTrainingParameters.ComparerBase.CompareMinError
getset

Whether the min error in convergence table is used for comparison when errors are compared. If false then mean value of last errors is used.

int IG.Num.NeuralTrainingParameters.ComparerBase.NumLastErrors
getset

Number of last errors in convergence list for calculataing the mean value of error. The default valeu is 1 which represent the last error in the converhence list.

int IG.Num.NeuralTrainingParameters.ComparerBase.NumBundles
getset

Number of bundles where sorting of convergences stars. The default valeu is 0 which represent normal sorting after the training is done.

bool IG.Num.NeuralTrainingParameters.ComparerBase.CompareByTrainingRmsError
getset

Whether training RMS errors from convergence tavble are compared.

Referenced by IG.Num.TestTrainingParametersComparers.Test().

bool IG.Num.NeuralTrainingParameters.ComparerBase.CompareByTrainingMaxError
getset

Whether Maximal absolute training errors from convergence tavble are compared.

bool IG.Num.NeuralTrainingParameters.ComparerBase.CompareByVerificationRmsError
getset

Whether verification RMS errors from convergence tavble are compared.

bool IG.Num.NeuralTrainingParameters.ComparerBase.CompareByVerificationMaxError
getset

Whether Maximal absolute verification errors from convergence tavble are compared.


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