Class DiceLoss<T>

Namespace

AiDotNet.LossFunctions

Assembly

AiDotNet.dll

Implements the Dice loss function, commonly used for image segmentation tasks.

public class DiceLoss<T> : LossFunctionBase<T>, ILossFunction<T>

Type Parameters

T

The numeric type used for calculations (e.g., float, double).

Inheritance

object

LossFunctionBase<T>

DiceLoss<T>

Implements

ILossFunction<T>

Inherited Members

LossFunctionBase<T>.NumOps

LossFunctionBase<T>.Engine

LossFunctionBase<T>.CalculateLossAndGradientGpu(IGpuTensor<T>, IGpuTensor<T>)

LossFunctionBase<T>.ValidateVectorLengths(Vector<T>, Vector<T>)

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Extension Methods

LossFunctionExtensions.ComputeGradient<T>(ILossFunction<T>, Tensor<T>, Tensor<T>)

LossFunctionExtensions.ComputeLoss<T>(ILossFunction<T>, Tensor<T>, Tensor<T>)

Remarks

For Beginners: Dice loss measures the overlap between predicted and actual segments in an image. It's based on the Dice coefficient (also known as F1 score), which is a statistical measure of similarity.

The formula is: DiceLoss = 1 - (2 * intersection) / (sum of predicted + sum of actual)

Where:

• intersection is the sum of element-wise multiplication of predicted and actual values
• A value of 0 means perfect overlap (ideal predictions)
• A value of 1 means no overlap at all (worst predictions)

Key properties:

• It's ideal for problems where the positive class (what you're trying to detect) is rare
• Handles imbalanced data better than cross-entropy in many cases
• Focuses on maximizing the overlap between predictions and ground truth
• Commonly used in medical image segmentation, satellite imagery, and other segmentation tasks

Unlike cross-entropy, which treats each pixel independently, Dice loss considers the global relationship between predicted and actual masks, which often leads to better segmentation results.

Constructors

DiceLoss()

Initializes a new instance of the DiceLoss class.

public DiceLoss()

Methods

CalculateDerivative(Vector<T>, Vector<T>)

Calculates the derivative of the Dice loss function.

public override Vector<T> CalculateDerivative(Vector<T> predicted, Vector<T> actual)

Parameters

predicted Vector<T>

The predicted values.

actual Vector<T>

The actual (target) values.

Returns

Vector<T>

A vector containing the derivatives of Dice loss for each prediction.

CalculateLoss(Vector<T>, Vector<T>)

Calculates the Dice loss between predicted and actual values.

public override T CalculateLoss(Vector<T> predicted, Vector<T> actual)

Parameters

predicted Vector<T>

The predicted values (typically probabilities between 0 and 1).

actual Vector<T>

The actual (target) values (typically 0 or 1).

Returns

T

The Dice loss value.