AdversarialAttackBase<T, TInput, TOutput>

Base Classes · AiDotNet.AdversarialRobustness.Attacks

Base class for adversarial attack implementations.

Implements: IAdversarialAttack<T, TInput, TOutput>, IModelShape

Known derived types: AutoAttack<T, TInput, TOutput>, CWAttack<T, TInput, TOutput>, FGSMAttack<T, TInput, TOutput>, PGDAttack<T, TInput, TOutput>, ViTAdversarialAttack<T, TInput, TOutput>

For Beginners

This provides AI safety functionality. Default values follow the original paper settings.

Methods

Method Summary
AdversarialAttackOptions<T> GetOptions()
DynamicShapeInfo GetDynamicShapeInfo()
TInput CalculatePerturbation(TInput original, TInput adversarial)
TInput GenerateAdversarialExample(TInput input, TOutput trueLabel, IFullModel<T, TInput, TOutput> targetModel)
TInput[] GenerateAdversarialBatch(TInput[] inputs, TOutput[] trueLabels, IFullModel<T, TInput, TOutput> targetModel)
byte[] Serialize()
int[] GetInputShape() Returns the input shape for this attack configuration.
int[] GetOutputShape() Returns the output shape for this attack configuration.
void Deserialize(byte[] data)
void LoadModel(string filePath)
void Reset()
void SaveModel(string filePath)