adversarial attacks on image processing

need to write about the following threat model attacks (3 – 4 lines for every attack with the mathematical formulas), also list all the notations used in formulas with a brief descriiption.

You can refer to the attached paper take the threat model sections write till other attacks (Please use another notation for mathematical formulas). In this paper author has explained in detailed, but you just write about 3 to 4 lines about the attack with mathematically formulas.

Attacks are following
– Fast Gradient Sign Method (FGSM)
– L2DeepFoolAttack: A simple and fast gradient-based adversarial attack.
– DDNAttack: The Decoupled Direction and Norm L2 adversarial attack.
– InversionAttack: Creates “negative images” by inverting the pixel values.
– NewtonFoolAttack: Implementation of the NewtonFool Attack.
– LinfPGD: Linf Projected Gradient Descent
– SaltAndPepperNoiseAttack: Increases the amount of salt and pepper noise until the input is misclassified.

Last Completed Projects