API Reference - CostFunctions

Constructors

FastBinaryCrossEntropy

CostFunction.FastBinaryCrossEntropy{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

FastCategoricalCrossEntropy

CostFunction.FastCategoricalCrossEntropy{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

FastFocalLoss

CostFunction.FastFocalLoss{generatedLabelTensor: tensor, labelTensor: tensor, alpha: number, gamma: number}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

• alpha: The weighting factor used to deal with class imbalance. Must be between 0 and 1. [Default: 0.25]

• gamma: An adjustable focusing parameter. Must be a positive value. [Default: 2]

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

FastMeanAbsoluteError

CostFunction.FastMeanAbsoluteError{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

FastMeanSquaredError

CostFunction.FastMeanSquaredError{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

BinaryCrossEntropy

CostFunction.BinaryCrossEntropy{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

CategoricalCrossEntropy

CostFunction.CategoricalCrossEntropy{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

FocalLoss

CostFunction.FocalLoss{generatedLabelTensor: tensor, labelTensor: tensor, alpha: number, gamma: number}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

• alpha: The weighting factor used to deal with class imbalance. Must be between 0 and 1. [Default: 0.25]

• gamma: An adjustable focusing parameter. Must be a positive value. [Default: 2]

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

MeanAbsoluteError

CostFunction.MeanAbsoluteError{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.

MeanSquaredError

CostFunction.MeanSquaredError{generatedLabelTensor: tensor, labelTensor: tensor}: AutomaticDifferentiationTensor

Parameters:

• generatedLabelTensor: The tensor that is generated by a model.

• labelTensor: The tensor that will be used as a target by a model.

Returns:

• AutomaticDifferentiationTensor: The automatic differentiation tensor that is created as a result of calling this function.