Page 230 - AI Computer 10

5. Differentiate between precision and recall.

6. What key insights does a confusion matrix provide in model evaluation?
7. What are False Positives and False Negatives?
8. What is the role of training and testing dataset in model evaluation?
9. Write three points of ethical concerns related to evaluation of AI models.

10. Define the following terms in context of AI model evaluation:
(a) Predicti on (b) Reality

(c) Accuracy (d) Error
F. Long answer type questions.
1. Explain the concept of a False Negative and its implications in evaluating the performance of an AI model,
giving a relevant example.
2. Describe F1 Score. How is it calculated? Discuss why it is considered a balanced metric in model evaluation.

3. How does the concept of Recall contribute to understanding the effectiveness of an AI model, and in
what types of applications is it particularly crucial?
4. Describe various ethical concerns related to AI model evaluation techniques and metrics.
5. Describe the Confusion Matrix and its components.

6. Describe the various metrics involved in AI model evaluation.
G. Assertion and Reason-based questions.
In the following questions, a statement of Assertion (A) is followed by a statement of Reason (R). Study both
the statements and give the answer as:

(a) if both Assertion and Reason are correct and Reason is the correct explanation of Assertion.
(b) if both Assertion and Reason are correct but Reason is not the correct explanation of Assertion.
(c) if Assertion is True but Reason is False.

(d) if Assertion is False but Reason is True.
1. Assertion (A): Model evaluation helps in detecting overfitting.
Reason (R): Overfitting occurs when a model performs well on training data but poorly on test data.
2. Assertion (A): The train-test split method ensures that a model generalises well to new data.

Reason (R): A model should be evaluated using the same data on which it was trained.
3. Assertion (A): A model with very high training accuracy is always the best model.

Reason (R): High training accuracy may indicate overfitting, leading to poor generalisation on test data.
4. Assertion (A): Ethical concerns are not relevant in AI model evaluation.
Reason (R): Bias in model evaluation can lead to unfair and inaccurate predictions, affecting real-world
decisions.

AI Fun Zone 21st
Century Project-based Learning
Skills
Skills
Skills
1. Prepare a presentation on the various AI model evaluation metrics, giving the formula for each.
2. Prepare a chart paper listing various ethical concerns around AI model evaluation.

96
96