eVision.EasyDeepLearning.EClassificationMetrics Class Reference

Collection of metrics used to evaluate the state of an EClassifier. More...

Inheritance diagram for eVision.EasyDeepLearning.EClassificationMetrics:

Collaboration diagram for eVision.EasyDeepLearning.EClassificationMetrics:

Public Member Functions
	EClassificationMetrics ()
	Constructs an EClassificationMetrics object.
	EClassificationMetrics (eVision.EasyDeepLearning.EClassificationMetrics other)
	Copy constructor of an EClassificationMetrics object.
void	Load (string path)
	Loads a classification metric. The given ESerializer must have been created for reading.
void	Load (eVision.ESerializer serializer)
	Loads a classification metric. The given ESerializer must have been created for reading.
void	Save (string path)
	Saves a classification metric. The given ESerializer must have been created for writing.
void	Save (eVision.ESerializer serializer)
	Saves a classification metric. The given ESerializer must have been created for writing.
bool	IsValid ()
	Indicates whether the object contains at least one classification result.
void	AddResult (eVision.EasyDeepLearning.EClassificationResult result, string groundtruthLabel)
	Adds the given result with the corresponding ground truth label to the metrics.
void	AddMetrics (eVision.EasyDeepLearning.EClassificationMetrics other)
	Adds the other metrics to the current metrics of this object.
float	GetWeightedAccuracy (float[] weights)
	The label weighted accuracy.
float	GetWeightedAccuracy (eVision.EasyDeepLearning.EDataset dataset)
	The label weighted accuracy.
float	GetLabelAccuracy (string label)
	The accuracy of the classifier for a given label.
float	GetWeightedError (float[] weights)
	The label weighted error.
float	GetWeightedError (eVision.EasyDeepLearning.EDataset dataset)
	The label weighted error.
float	GetLabelError (string label)
	The error of the classifier for a given label.
bool	CanComputeWeightedError ()
	Whether the object can be used to get weighted, balanced, and label errors.
uint	GetConfusion (string trueClass, string predictedClass)
	Confusion value of one label with another.

Properties
float	Accuracy [get]
	The accuracy of the classifier.
float	BalancedAccuracy [get]
	The balanced accuracy.
float	Error [get]
	The error of the classifier.
float	BalancedError [get]
	The balanced error.

Detailed Description

Collection of metrics used to evaluate the state of an EClassifier.

A metric is a value summarizing a collection of classification results (EClassificationResult). New results can be added to the object individually with EClassificationMetrics.AddResult or collectively with EClassificationMetrics.AddMetrics.

EClassificationMetrics contains the following metrics:

• the accuracy (see EClassificationMetrics.Accuracy)
• the error (see EClassificationMetrics.Error)

License(s): BasicTypes

Constructor & Destructor Documentation

EClassificationMetrics() [1/2]

eVision.EasyDeepLearning.EClassificationMetrics.EClassificationMetrics

(

)

Constructs an EClassificationMetrics object.

EClassificationMetrics() [2/2]

eVision.EasyDeepLearning.EClassificationMetrics.EClassificationMetrics

(

eVision.EasyDeepLearning.EClassificationMetrics

other

)

Copy constructor of an EClassificationMetrics object.

Parameters

other

Reference to the EClassificationMetrics object that should be copied

Member Function Documentation

AddMetrics()

void eVision.EasyDeepLearning.EClassificationMetrics.AddMetrics

(

eVision.EasyDeepLearning.EClassificationMetrics

other

)

Adds the other metrics to the current metrics of this object.

Parameters

other

Classification metrics

AddResult()

void eVision.EasyDeepLearning.EClassificationMetrics.AddResult	(	eVision.EasyDeepLearning.EClassificationResult	result,
		string	groundtruthLabel )

Adds the given result with the corresponding ground truth label to the metrics.

Parameters

result	A reference to an EClassificationResult object.
groundtruthLabel	The ground truth label corresponding to the result

CanComputeWeightedError()

bool eVision.EasyDeepLearning.EClassificationMetrics.CanComputeWeightedError

(

)

Whether the object can be used to get weighted, balanced, and label errors.

GetConfusion()

uint eVision.EasyDeepLearning.EClassificationMetrics.GetConfusion	(	string	trueClass,
		string	predictedClass )

Confusion value of one label with another.

The confusion value of a label with another is the number of images belonging to this label that are classified as belonging to the other label.

Parameters

trueClass	The label for which to obtain the confusion value
predictedClass	The label with which there is a confusion

GetLabelAccuracy()

float eVision.EasyDeepLearning.EClassificationMetrics.GetLabelAccuracy

(

string

label

)

The accuracy of the classifier for a given label.

Parameters

label

The label of which we want the accuracy

The label accuracy is the number of images of a given label that were correctly classified (also called the true positives) over the total number of images of that label that was used to evaluate the classifier.

If there is no results for the given label, the method will throw an exception.

GetLabelError()

float eVision.EasyDeepLearning.EClassificationMetrics.GetLabelError

(

string

label

)

The error of the classifier for a given label.

Parameters

label

The label from which to get the error.

The label error corresponds to the error only for images of the given label.

If there is no results for the given label, the method will throw an exception. If EClassificationMetrics.CanComputeWeightedError is false, this method is an alias for EClassificationMetrics.Error.

GetWeightedAccuracy() [1/2]

float eVision.EasyDeepLearning.EClassificationMetrics.GetWeightedAccuracy

(

eVision.EasyDeepLearning.EDataset

dataset

)

The label weighted accuracy.

Parameters

dataset

Dataset from which to use the label weights

The weighted accuracy is the weighted average of the label accuracies (see EClassificationMetrics.GetLabelAccuracy). If there is no results for a given label, it will be ignored in the weighted accuracy.

GetWeightedAccuracy() [2/2]

float eVision.EasyDeepLearning.EClassificationMetrics.GetWeightedAccuracy

(

float[]

weights

)

The label weighted accuracy.

Parameters

weights

Array of label weights for the labels of the classifier (see EDeepLearningTool.GetLabel)

The weighted accuracy is the weighted average of the label accuracies (see EClassificationMetrics.GetLabelAccuracy). If there is no results for a given label, it will be ignored in the weighted accuracy.

GetWeightedError() [1/2]

float eVision.EasyDeepLearning.EClassificationMetrics.GetWeightedError

(

eVision.EasyDeepLearning.EDataset

dataset

)

The label weighted error.

Parameters

dataset

Dataset from which to use the label weights

The weighted error is the weighted average of the label errors (see EClassificationMetrics.GetLabelError).

If there isn't any result for a given label, it will be ignored in the weighted error.

If EClassificationMetrics.CanComputeWeightedError is false, this method is an alias for EClassificationMetrics.Error.

GetWeightedError() [2/2]

float eVision.EasyDeepLearning.EClassificationMetrics.GetWeightedError

(

float[]

weights

)

The label weighted error.

Parameters

weights

Array of label weights for the labels of the classifier (see EDeepLearningTool.GetLabel)

The weighted error is the weighted average of the label errors (see EClassificationMetrics.GetLabelError).

If there isn't any result for a given label, it will be ignored in the weighted error.

If EClassificationMetrics.CanComputeWeightedError is false, this method is an alias for EClassificationMetrics.Error.

IsValid()

bool eVision.EasyDeepLearning.EClassificationMetrics.IsValid

(

)

Indicates whether the object contains at least one classification result.

Load() [1/2]

void eVision.EasyDeepLearning.EClassificationMetrics.Load

(

eVision.ESerializer

serializer

)

Loads a classification metric. The given ESerializer must have been created for reading.

Parameters

serializer

The serializer.

Load() [2/2]

void eVision.EasyDeepLearning.EClassificationMetrics.Load

(

string

path

)

Loads a classification metric. The given ESerializer must have been created for reading.

Parameters

path	The file path.

Save() [1/2]

void eVision.EasyDeepLearning.EClassificationMetrics.Save

(

eVision.ESerializer

serializer

)

Saves a classification metric. The given ESerializer must have been created for writing.

Parameters

serializer

The ESerializer object that is written to.

Save() [2/2]

void eVision.EasyDeepLearning.EClassificationMetrics.Save

(

string

path

)

Saves a classification metric. The given ESerializer must have been created for writing.

Parameters

path	The file path.

Property Documentation

Accuracy

float eVision.EasyDeepLearning.EClassificationMetrics.Accuracy

get

The accuracy of the classifier.

The accuracy is the number of images that were correctly classified (also called the true positives) over the total number of images that was used to evaluate the classifier.

BalancedAccuracy

float eVision.EasyDeepLearning.EClassificationMetrics.BalancedAccuracy

get

The balanced accuracy.

The balanced accuracy is the label weighted accuracy with the same weight for each labels.

BalancedError

float eVision.EasyDeepLearning.EClassificationMetrics.BalancedError

get

The balanced error.

The balanced error is the label weighted error with the same weight for each labels.

If EClassificationMetrics.CanComputeWeightedError is false, this method is an alias for EClassificationMetrics.Error.

Error

float eVision.EasyDeepLearning.EClassificationMetrics.Error

get

The error of the classifier.

The error, which is also called the loss, is the quantity that is minimized during the training of the deep neural network. For classification, the error is the cross-entropy.