- A
Use a post-processing step to blur or distort generated images.
Why wrong: This reduces quality and does not address the root cause of memorization.
- B
Re-tune the model using differential privacy (DP-SGD) to prevent memorization of individual examples.
Differential privacy limits what the model can learn about specific training examples, reducing memorization.
- C
Increase the size of the training dataset by adding more synthetic images.
Why wrong: Adding more data might not solve the memorization issue if the model overfits to specific examples.
- D
Apply stricter output safety filters to block images that look like any known patient.
Why wrong: Safety filters are not designed to detect similarity to training data.
Question 274 of 997
Differential Privacy for Imagen Fine-Tuning
This Generative AI Leader practice question tests your understanding of google cloud's generative ai offerings. The scenario asks you to isolate a root cause — eliminate options that address a different problem before choosing. After answering, compare your reasoning against the explanation and wrong-answer breakdown below. Once you have made your selection, read the full explanation to reinforce the concept and understand why each distractor is designed to mislead on exam day.
A healthcare startup is using Vertex AI Imagen to generate synthetic medical images for training a diagnostic model. The images must comply with HIPAA regulations and cannot contain any real patient data. The team fine-tuned Imagen on a dataset of de-identified medical scans. However, during testing, they notice that some generated images closely resemble specific patients from the original dataset, even though the dataset was de-identified. They suspect that the model memorized some training examples. The team needs to address this issue without losing image quality. They have access to the original training data and Vertex AI tools. What action should they take?
Quick Answer
The correct action is to re-tune the model using differential privacy (DP-SGD) to prevent memorization of individual examples. This technique, specifically Differentially Private Stochastic Gradient Descent, works by injecting calibrated noise into the gradient updates during fine-tuning, which mathematically bounds the model’s ability to memorize any single patient record from the training set. On the Google Cloud Generative AI Leader exam, this scenario tests your understanding of how to balance privacy compliance with generative output quality, especially under HIPAA constraints for Vertex AI Imagen. A common trap is confusing output-level safety filters with training-time privacy guarantees—filters only catch obvious violations, not subtle patient recall. Remember the key distinction: differential privacy protects during training, while safety filters protect after generation. Memory tip: think of DP-SGD as adding “statistical static” to the learning process so the model sees the forest, not the individual trees.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Re-tune the model using differential privacy (DP-SGD) to prevent memorization of individual examples.
Option B is correct because differential privacy (DP-SGD) during fine-tuning adds calibrated noise to the gradient updates, which mathematically bounds the model's ability to memorize any single training example. This directly addresses the memorization issue while preserving the utility of the generated images, as the noise is carefully controlled to maintain overall image quality. Vertex AI supports DP-SGD through its custom training infrastructure, making it a practical choice for HIPAA-compliant medical imaging.
Key principle: Answer the scenario, not the keyword: identify the specific constraint before choosing the most familiar-sounding option.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Use a post-processing step to blur or distort generated images.
Why it's wrong here
This reduces quality and does not address the root cause of memorization.
- ✓
Re-tune the model using differential privacy (DP-SGD) to prevent memorization of individual examples.
Why this is correct
Differential privacy limits what the model can learn about specific training examples, reducing memorization.
Related concept
Read the scenario before looking for a memorised answer.
- ✗
Increase the size of the training dataset by adding more synthetic images.
Why it's wrong here
Adding more data might not solve the memorization issue if the model overfits to specific examples.
- ✗
Apply stricter output safety filters to block images that look like any known patient.
Why it's wrong here
Safety filters are not designed to detect similarity to training data.
Common exam traps
Common exam trap: answer the scenario, not the keyword
Google often tests the misconception that post-processing or filtering can solve memorization, when in fact the root cause is in the training algorithm itself, requiring a privacy-preserving technique like differential privacy.
Trap categories for this question
Similar concept trap
Safety filters are not designed to detect similarity to training data.
Detailed technical explanation
How to think about this question
Differential privacy (DP-SGD) works by clipping gradients to a fixed norm and adding Gaussian noise proportional to the privacy budget (epsilon). In medical imaging, a typical epsilon value of 1–10 can prevent memorization while retaining diagnostic utility. A real-world scenario is the use of DP-SGD in federated learning for hospital networks, where patient data must remain private across institutions.
KKey Concepts to Remember
- Read the scenario before looking for a memorised answer.
- Find the constraint that changes the correct option.
- Eliminate answers that are true in general but not in this case.
TExam Day Tips
- Watch for words such as best, first, most likely and least administrative effort.
- Review why wrong options are wrong, not only why the correct option is correct.
Key takeaway
Answer the scenario, not the keyword: identify the specific constraint before choosing the most familiar-sounding option.
Real-world example
How this comes up in practice
A company's IT admin needs to give a contractor read-only access to production logs without sharing account credentials. Using role-based access control (RBAC) and temporary scoped permissions — not a permanent shared password — is the correct pattern. Questions like this test whether you can apply least-privilege access across cloud identity services.
What to study next
Got this wrong? Here's your next step.
Identify which exam domain this question belongs to, review the core concept, then practise similar questions from the same domain.
- →
Google Cloud's Generative AI Offerings — study guide chapter
Learn the concepts, then practise the questions
- →
Google Cloud's Generative AI Offerings practice questions
Targeted practice on this topic area only
- →
All Generative AI Leader questions
997 questions across all exam domains
- →
Google Cloud Generative AI Leader Generative AI Leader study guide
Full concept coverage aligned to exam objectives
- →
Generative AI Leader practice test guide
How to use practice tests most effectively before exam day
Related practice questions
Related Generative AI Leader practice-question pages
Use these pages to review the topic behind this question. This is how one missed question becomes focused revision.
Fundamentals of Generative AI practice questions
Practise Generative AI Leader questions linked to Fundamentals of Generative AI.
Business Strategies for Generative AI Solutions practice questions
Practise Generative AI Leader questions linked to Business Strategies for Generative AI Solutions.
Generative AI Concepts and Technologies practice questions
Practise Generative AI Leader questions linked to Generative AI Concepts and Technologies.
Google AI Ecosystem and Strategy practice questions
Practise Generative AI Leader questions linked to Google AI Ecosystem and Strategy.
Responsible AI and Data Governance practice questions
Practise Generative AI Leader questions linked to Responsible AI and Data Governance.
Google Cloud's Generative AI Offerings practice questions
Practise Generative AI Leader questions linked to Google Cloud's Generative AI Offerings.
Techniques to Improve Generative AI Model Output practice questions
Practise Generative AI Leader questions linked to Techniques to Improve Generative AI Model Output.
Applying Generative AI in Business practice questions
Practise Generative AI Leader questions linked to Applying Generative AI in Business.
Generative AI Leader fundamentals practice questions
Practise Generative AI Leader questions linked to Generative AI Leader fundamentals.
Generative AI Leader scenario practice questions
Practise Generative AI Leader questions linked to Generative AI Leader scenario.
Generative AI Leader troubleshooting practice questions
Practise Generative AI Leader questions linked to Generative AI Leader troubleshooting.
Practice this exam
Start a free Generative AI Leader practice session
Short sessions build daily habit. Longer sessions build exam-day stamina. Try a timed session to simulate real conditions.
FAQ
Questions learners often ask
What does this Generative AI Leader question test?
Google Cloud's Generative AI Offerings — This question tests Google Cloud's Generative AI Offerings — Read the scenario before looking for a memorised answer..
What is the correct answer to this question?
The correct answer is: Re-tune the model using differential privacy (DP-SGD) to prevent memorization of individual examples. — Option B is correct because differential privacy (DP-SGD) during fine-tuning adds calibrated noise to the gradient updates, which mathematically bounds the model's ability to memorize any single training example. This directly addresses the memorization issue while preserving the utility of the generated images, as the noise is carefully controlled to maintain overall image quality. Vertex AI supports DP-SGD through its custom training infrastructure, making it a practical choice for HIPAA-compliant medical imaging.
What should I do if I get this Generative AI Leader question wrong?
Identify which exam domain this question belongs to, review the core concept, then practise similar questions from the same domain.
What is the key concept behind this question?
Read the scenario before looking for a memorised answer.
About these practice questions
Courseiva creates original exam-style practice questions with explanations and wrong-answer analysis. It does not publish real exam questions, exam dumps, or protected exam content. Learn why practice questions differ from exam dumps →
Keep practising
More Generative AI Leader practice questions
- A data scientist is trying to get online predictions from a Vertex AI endpoint but receives the error shown. What is the…
- A data scientist notices that a text generation model deployed on Vertex AI returns repetitive outputs after a few turns…
- A company is deploying a generative AI model for medical diagnosis support. Which THREE considerations are critical for…
- Which THREE considerations are critical when deploying a generative AI model using Vertex AI Endpoints for a latency-sen…
- A company is deploying a generative AI model for customer support. They want to reduce hallucinations while maintaining…
- Which TWO techniques are commonly used to control the style and tone of a generative model's output?
Last reviewed: Jul 4, 2026
This Generative AI Leader practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the Generative AI Leader exam.
Question Discussion
Share a tip, memory trick, or ask about the reasoning behind this question. Do not post real exam questions, leaked content, braindumps, or copyrighted exam material. Comments are moderated and may be removed without notice.
Sign in to join the discussion.