ISTQB AI Neural Network Testing Knowledge

1. True or False: Regression testing is not applicable to AI and machine learning systems.

True

False

2. True or False: Once an AI model passes initial testing, no further testing is needed during its operational lifetime.

True

False

3. What does 'poisoning attack' testing evaluate in AI security?

Model vulnerability to malicious training data injection

The nutritional value of food models

The speed of data transmission

The color accuracy of image recognition

4. Which factor is most critical when designing test data for neural network validation?

Ensuring data distribution matches production scenarios and includes edge cases

Using the smallest possible dataset

Creating identical inputs and outputs

Avoiding any preprocessing of data

5. True or False: Explainability and interpretability mean exactly the same thing in AI testing contexts.

True

False

6. In ISTQB AI testing, which of these is a valid approach to address the test oracle problem?

Using metamorphic relations and human expert review

Avoiding testing altogether

Testing only with random inputs

Assuming all outputs are correct

7. What is a 'test oracle problem' in AI testing?

Difficulty determining whether an AI output is correct when no ground truth exists

A tool that automatically generates all test cases

The phase where developers write code

A method to increase model accuracy instantly

8. True or False: Fairness testing ensures AI models produce equitable outcomes across different demographic groups.

True

False

9. Which testing technique is most appropriate for detecting performance degradation in deployed AI models?

Continuous monitoring and model drift detection

Testing only before deployment

Manual user feedback exclusively

Ignoring production data entirely

10. In neural network testing, what does 'neuron coverage' measure?

The percentage of neurons activated by test inputs during execution

The total number of neurons in the network

The accuracy of predictions only

The physical size of the network

11. What is the primary purpose of metamorphic testing in AI systems?

To verify outputs follow mathematical relationships without knowing expected results

To test only the neural network architecture

To replace traditional unit testing entirely

To measure computational speed only

12. What is 'model drift' in AI systems?

Degradation of model performance over time due to changes in data distribution or environment

The initial training phase of a neural network

A feature that improves accuracy automatically

A type of data preprocessing technique

13. Which of the following is essential for testing AI model robustness?

Testing with out-of-distribution and edge-case data

Testing only with the training dataset

Ignoring data preprocessing steps

Using a single test case repeatedly

14. True or False: Data quality has minimal impact on neural network model testing outcomes.

True

False

15. In the context of ISTQB AI testing, what does 'model explainability' testing evaluate?

Whether stakeholders can understand and interpret model decisions

Only the mathematical accuracy of predictions

The visual appearance of the application

The cost of hosting the model

16. What is the primary goal of bias testing in AI systems?

To identify and mitigate unfair or discriminatory predictions across protected attributes

To ensure the model is as complex as possible

To increase model training time

To reduce the number of input features

17. Which testing approach uses inputs deliberately designed to fool a neural network?

Adversarial testing

Regression testing

Smoke testing

Boundary value analysis

18. True or False: Test coverage metrics for traditional software apply directly to neural networks without modification.

True

False

19. In AI testing, what does 'adversarial robustness' refer to?

A model's ability to maintain accuracy when exposed to intentionally crafted adversarial inputs

The speed at which a model trains

The number of parameters in the neural network

The color scheme of the user interface

20. Which of the following is a key challenge in testing neural networks?

The lack of clear test oracles for prediction validation

Neural networks always produce identical outputs

Input data is always perfectly labeled

Model complexity never affects testing difficulty

ISTQB AI Testing Neural Network Testing Quiz

1. True or False: Regression testing is not applicable to AI and machine learning systems.

2.

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

2. True or False: Once an AI model passes initial testing, no further testing is needed during its operational lifetime.

3. What does 'poisoning attack' testing evaluate in AI security?

4. Which factor is most critical when designing test data for neural network validation?

5. True or False: Explainability and interpretability mean exactly the same thing in AI testing contexts.

6. In ISTQB AI testing, which of these is a valid approach to address the test oracle problem?

7. What is a 'test oracle problem' in AI testing?

8. True or False: Fairness testing ensures AI models produce equitable outcomes across different demographic groups.

9. Which testing technique is most appropriate for detecting performance degradation in deployed AI models?

10. In neural network testing, what does 'neuron coverage' measure?

11. What is the primary purpose of metamorphic testing in AI systems?

12. What is 'model drift' in AI systems?

13. Which of the following is essential for testing AI model robustness?

14. True or False: Data quality has minimal impact on neural network model testing outcomes.

15. In the context of ISTQB AI testing, what does 'model explainability' testing evaluate?

16. What is the primary goal of bias testing in AI systems?

17. Which testing approach uses inputs deliberately designed to fool a neural network?

18. True or False: Test coverage metrics for traditional software apply directly to neural networks without modification.

19. In AI testing, what does 'adversarial robustness' refer to?

20. Which of the following is a key challenge in testing neural networks?