Mitigate Adversarial Risks CY0-001 Practice Questions (Page 6)

Questions 26–30

1. 26application · medium

   An e-commerce company uses a machine learning model to detect fraudulent product reviews. The security team notices a sudden increase in false positives, where legitimate reviews are flagged as fraudulent. They suspect an adversarial attack. Which monitoring technique would best help them identify the cause?

   • AReview the model's training data to see if any new examples were added recently.
   • BTrack the model's prediction confidence scores and compare them to historical baselines.
   • CUse a different model algorithm that is less sensitive to input variations.
   • DIncrease the model's retraining frequency to adapt to new patterns.
   Reveal Solution Discussion Select an answer first
2. 27application · medium

   A hospital uses an AI model to predict patient readmission risk. The compliance team requires that the model's decisions be explainable to auditors. Which technique would best meet this requirement?

   • ARetrain the model with more data to improve generalization.
   • BUse a more complex model to improve prediction accuracy.
   • CUse LIME (Local Interpretable Model-agnostic Explanations) to generate explanations for individual predictions.
   • DMonitor the model's performance on a weekly basis.
   Reveal Solution Discussion Select an answer first
4. 28application · medium

   An autonomous vehicle company wants to make its object detection model more robust against adversarial stickers that can be placed on stop signs to confuse the model. Which technique is most likely to improve the model's robustness?

   • ADeploy the model on more powerful hardware to process images faster.
   • BAdd more layers to the neural network to increase its capacity.
   • CUse a higher-resolution camera to capture more details of the stop sign.
   • DTrain the model with a diverse set of adversarial examples, including sticker-like perturbations.
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A government agency is deploying an AI system for visa application screening. During threat modeling, they identify that an adversary could submit crafted applications to influence the model's decisions. Which threat modeling approach would best help them systematically identify and prioritize such risks?

   • AReview the model's accuracy on a validation dataset.
   • BConduct a penetration test on the AI model's API to find vulnerabilities.
   • CUse the STRIDE methodology to categorize threats such as spoofing, tampering, and denial of service.
   • DImplement a firewall to protect the model's infrastructure.
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A social media platform uses an AI model to filter hate speech. The security team discovers that an attacker has been submitting large volumes of benign posts that are labeled as hate speech to manipulate the model's training data. What is the most effective way to ensure data integrity in this scenario?

   • AUse a larger training dataset to dilute the impact of the poisoned data.
   • BIncrease the model's complexity to better distinguish between benign and hate speech.
   • CRestrict access to the model's API to prevent attackers from submitting data.
   • DImplement a data provenance system that tracks the origin and labeling of all training data.
   Reveal Solution Discussion Select an answer first