Differential Privacy & Federated Learning Quiz

1. Federated learning requires raw user data to be sent to a central server.

True

False

2. What is the main purpose of adding noise in differential privacy mechanisms?

To obscure individual records while maintaining statistical utility

To encrypt sensitive information

To reduce model accuracy intentionally

To prevent unauthorized access

3. Differential privacy guarantees can be formally proven mathematically.

True

False

4. Which algorithm is commonly used for secure aggregation in federated learning?

Secure multi-party computation

Simple averaging

Random sampling

Majority voting

Submit

6. What is 'differential privacy composition'?

Combining multiple DP mechanisms and their privacy budgets

Merging datasets from different sources

Stacking neural network layers

Aggregating model predictions

7. In federated learning, model updates are aggregated using which method?

Secure aggregation at a central coordinator

Direct data sharing between nodes

Public key exchange

Password-protected uploads

8. Differential privacy is only applicable to structured data.

True

False

9. Which scenario best demonstrates federated learning application?

Multiple hospitals training models on local patient data without sharing records

Centralizing all medical records in one database

Encrypting data before sending to cloud servers

Using public datasets for model training

10. The Gaussian mechanism in differential privacy adds noise from which distribution?

Gaussian (normal) distribution

Uniform distribution

Exponential distribution

Poisson distribution

11. What is differential privacy?

A technique adding noise to data to protect individual privacy

A method of encrypting all data in transit

A process of removing personally identifiable information

A firewall rule that blocks unauthorized access

12. What does 'local differential privacy' mean?

Privacy protection at individual user level before data aggregation

Privacy limited to local network only

Privacy applied only during model training

Privacy enforced by local administrators

13. Differential privacy can be applied to which machine learning operations?

Training, inference, and data publishing

Only model training

Only data collection

Only model deployment

14. What is a key challenge in federated learning?

Communication overhead between nodes

Lack of data standardization

Simplified model deployment

Reduced privacy concerns

Submit

16. What is a primary advantage of federated learning over centralized learning?

Reduced data exposure and improved privacy

Faster training on single machines

Simplified model architecture

Lower computational requirements

17. Federated learning reduces the risk of data breaches by avoiding centralized data collection.

True

False

18. Which mechanism is commonly used to implement differential privacy?

Laplace mechanism adding calibrated noise

Hash-based authentication

Public key infrastructure

Role-based access control

19. What does epsilon (ε) represent in differential privacy?

The privacy budget or level of privacy protection

The encryption strength

The model accuracy threshold

The number of training iterations

20. In federated learning, where is the model training performed?

Only on a central server

On distributed edge devices or local nodes

In cloud storage exclusively

On encrypted external drives

SecAI+ Differential Privacy and Federated Learning Quiz

1. Federated learning requires raw user data to be sent to a central server.

2.

What first name or nickname would you like us to use?

2. What is the main purpose of adding noise in differential privacy mechanisms?

3. Differential privacy guarantees can be formally proven mathematically.

4. Which algorithm is commonly used for secure aggregation in federated learning?

5. Federated learning is particularly useful in scenarios requiring data ______.

6. What is 'differential privacy composition'?

7. In federated learning, model updates are aggregated using which method?

8. Differential privacy is only applicable to structured data.

9. Which scenario best demonstrates federated learning application?

10. The Gaussian mechanism in differential privacy adds noise from which distribution?

11. What is differential privacy?

12. What does 'local differential privacy' mean?

13. Differential privacy can be applied to which machine learning operations?

14. What is a key challenge in federated learning?

15. In differential privacy, decreasing epsilon value ______ privacy protection.

16. What is a primary advantage of federated learning over centralized learning?

17. Federated learning reduces the risk of data breaches by avoiding centralized data collection.

18. Which mechanism is commonly used to implement differential privacy?

19. What does epsilon (ε) represent in differential privacy?

20. In federated learning, where is the model training performed?