Module 3 Graded Quiz: Keras and Deep Learning Libraries :Introduction to Deep Learning & Neural Networks with Keras (IBM AI Engineering Professional Certificate) Answers 2025
1. Question 1 — Which statement is correct?
-
❌ TensorFlow is the cousin of Torch
-
❌ Google supports Keras and PyTorch
-
❌ PyTorch runs on top of TensorFlow
-
✅ Keras is a high-level API that facilitates fast development and prototyping
Explanation:
Keras is designed to be an easy, modular, high-level deep-learning API.
2. Question 2 — Beginner-friendly deep learning library?
-
❌ Theano
-
✅ Keras
-
❌ Scikit-learn
-
❌ TensorFlow
Explanation:
Keras is known for its simple syntax and ease of use.
3. Question 3 — Which model types does Keras support? (Select all)
-
❌ The Keras model
-
✅ The Subclassing model
-
✅ The Sequential model
-
❌ The Dense model
Explanation:
Keras supports Sequential, Functional, and Subclassing models.
4. Question 4 — Output activation for 10-class classification?
-
✅ Softmax
-
❌ ReLU
-
❌ Sigmoid
-
❌ Linear
Explanation:
Softmax outputs a probability distribution over multiple classes.
5. Question 5 — Best output activation for regression?
-
❌ Softmax
-
✅ Linear activation
-
❌ Sigmoid
-
❌ ReLU
Explanation:
Linear outputs any real value — best for continuous targets.
6. Question 6 — Best loss function for categorical outputs?
-
❌ MAE
-
✅ Categorical Cross-entropy
-
❌ Binary Cross-entropy
-
❌ MSE
Explanation:
Categorical cross-entropy evaluates multi-class predictions.
7. Question 7 — Best evaluation metric for continuous predictions?
-
✅ Mean Squared Error (MSE)
-
❌ F1-score
-
❌ Precision
-
❌ Accuracy
Explanation:
MSE and MAE are standard for regression performance.
8. Question 8 — Best compile settings for house-price regression?
-
❌ Categorical crossentropy + accuracy
-
❌ Binary crossentropy + precision
-
❌ Sparse categorical crossentropy + F1
-
✅ loss=’mean_squared_error’, optimizer=’adam’, metrics=[‘mean_absolute_error’]
Explanation:
Regression uses MSE/MAE and continuous output.
9. Question 9 — Best metric for a 3-class text classifier?
-
❌ Number of parameters
-
❌ Loss value only
-
✅ Accuracy score
-
❌ Training time
Explanation:
Accuracy gives a direct measure of correct classifications.
10. Question 10 — Why is Keras beginner-friendly?
-
❌ Least documentation
-
❌ Fastest execution
-
❌ Only for basic tasks
-
✅ Intuitive syntax & simplified model-building
Explanation:
Keras is known for clear, concise model definitions that beginners learn quickly.
🧾 Summary Table
| Q# | Correct Answer |
|---|---|
| 1 | Keras is a high-level API |
| 2 | Keras |
| 3 | Subclassing model, Sequential model |
| 4 | Softmax |
| 5 | Linear activation |
| 6 | Categorical cross-entropy |
| 7 | Mean Squared Error |
| 8 | MSE + Adam + MAE |
| 9 | Accuracy |
| 10 | Intuitive syntax & simple process |