LogLoss Class

Definition

The Log Loss, also known as the Cross Entropy Loss. It is commonly used in classification tasks.

public sealed class LogLoss : Microsoft.ML.Trainers.ILossFunction<float,float>, Microsoft.ML.Trainers.ISupportSdcaClassificationLoss
type LogLoss = class
    interface ISupportSdcaClassificationLoss
    interface ISupportSdcaLoss
    interface IScalarLoss
    interface ILossFunction<single, single>
    interface IClassificationLoss
Public NotInheritable Class LogLoss
Implements ILossFunction(Of Single, Single), ISupportSdcaClassificationLoss
Inheritance
LogLoss
Implements

Remarks

The Log Loss function is defined as:

$L(p(\hat{y}), y) = -y ln(\hat{y}) - (1 - y) ln(1 - \hat{y})$

where $\hat{y}$ is the predicted score, $p(\hat{y})$ is the probability of belonging to the positive class by applying a sigmoid function to the score, and $y \in \{0, 1\}$ is the true label.

Note that the labels used in this calculation are 0 and 1, unlike Hinge Loss and Exponential Loss, where the labels used are -1 and 1.

The Log Loss function provides a measure of how certain a classifier's predictions are, instead of just measuring how correct they are. For example, a predicted probability of 0.80 for a true label of 1 gets penalized more than a predicted probability of 0.99.

Constructors

LogLoss()

Methods

ComputeDualUpdateInvariant(Single)
Derivative(Single, Single)
DualLoss(Single, Single)
DualUpdate(Single, Single, Single, Single, Int32)
Loss(Single, Single)

Applies to