Which of the following is the MOST effective defense against cyberattacks that alter input data to avoid detection by the model?

The correct answer is B. Enhancing model robustness through adversarial training. The attack described - altering input data to evade model detection - is an adversarial evasion attack. The MOST effective countermeasure is adversarial training (B), which involves intentionally exposing the model to adversarial examples during training so it learns to…

AI Security Design and Implementation

Question

Options

AConducting periodic monitoring activities on the model's decisions
BEnhancing model robustness through adversarial training
CImplementing restricted access to the model's internal parameters
DApplying differential privacy controls on training datasets

How the community answered

(38 responses)

A
5% (2)
B
79% (30)
C
11% (4)
D
5% (2)

Explanation

The attack described - altering input data to evade model detection - is an adversarial evasion attack. The MOST effective countermeasure is adversarial training (B), which involves intentionally exposing the model to adversarial examples during training so it learns to correctly classify them. This directly hardens the model against this class of attack. Periodic monitoring (A) can detect problems after the fact but doesn't prevent evasion. Restricting access to internal parameters (C) is a model confidentiality control, not an evasion defense. Differential privacy (D) protects training data privacy, not model robustness against adversarial inputs at inference time.

Topics

#Adversarial attacks#Adversarial training#Model robustness#Input data integrity

Community Discussion

No community discussion yet for this question.

Full AAISM Practice