Class RobustnessMetrics<T>

Contains metrics for evaluating adversarial robustness of models.

public class RobustnessMetrics<T>

Inherited Members: object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Properties

Gets or sets additional evaluation metrics.

public Dictionary<string, double> AdditionalMetrics { get; set; }

Gets or sets the accuracy on adversarial examples.

public double AdversarialAccuracy { get; set; }

Gets or sets the attack success rate (percentage of successful attacks).

public double AttackSuccessRate { get; set; }

Gets or sets the average perturbation size needed to fool the model.

public double AveragePerturbationSize { get; set; }

Gets or sets the accuracy on clean (non-adversarial) examples.

public double CleanAccuracy { get; set; }

Gets or sets the robustness score (combines clean and adversarial accuracy).

public double RobustnessScore { get; set; }