Practice Questions

---

Question 1

Which characteristic best distinguishes deep learning from traditional machine learning techniques?

A. Deep learning always produces more accurate results
B. Deep learning uses rule-based logic
C. Deep learning uses neural networks with multiple layers
D. Deep learning does not require training data

Correct Answer: C

Explanation:
Deep learning is defined by the use of multi-layer (deep) neural networks, which allows the model to learn complex patterns. Accuracy is not guaranteed, and deep learning still requires training data.

---

Question 2

A data scientist is building a system to identify objects in photographs without manually defining features such as edges or shapes. Which approach best supports this requirement?

A. Linear regression
B. Decision trees
C. Deep learning
D. Rule-based classification

Correct Answer: C

Explanation:
Deep learning models automatically extract features from raw data, making them ideal for image recognition scenarios where manual feature engineering is difficult.

---

Question 3

Which type of data is deep learning particularly well suited to process?

A. Highly structured tabular data only
B. Unstructured data such as images and text
C. Small datasets with few attributes
D. Pre-aggregated numerical summaries

Correct Answer: B

Explanation:
Deep learning excels with unstructured data like images, audio, video, and natural language text — a key exam concept.

---

Question 4

Which scenario is the best example of a deep learning workload?

A. Predicting house prices using historical averages
B. Grouping customers by age and income
C. Translating spoken language into text
D. Calculating monthly sales totals

Correct Answer: C

Explanation:
Speech-to-text translation relies on deep neural networks trained on large datasets and is a classic deep learning use case.

---

Question 5

Why do deep learning models typically require large amounts of training data?

A. They rely on predefined rules
B. They use many layers with numerous parameters
C. They only work with structured data
D. They do not support feature reuse

Correct Answer: B

Explanation:
Deep learning models contain many parameters across multiple layers, which requires large datasets to train effectively and avoid overfitting.

---

Question 6

Which statement accurately describes feature engineering in deep learning?

A. Features must always be manually selected
B. Features are randomly generated
C. Features are automatically learned during training
D. Feature engineering is not possible

Correct Answer: C

Explanation:
A defining feature of deep learning is automatic feature extraction, reducing the need for manual feature engineering.

---

Question 7

Which Azure workload is most likely to use deep learning techniques?

A. Calculating averages in a SQL database
B. Performing rule-based fraud detection
C. Detecting faces in images
D. Sorting records by date

Correct Answer: C

Explanation:
Computer vision tasks such as face detection rely heavily on deep learning models.

---

Question 8

Compared to traditional machine learning models, deep learning models generally require:

A. Less computational power
B. No training data
C. More computational resources
D. Fewer model parameters

Correct Answer: C

Explanation:
Deep learning models are computationally intensive, often requiring GPUs and longer training times.

---

Question 9

Which statement is true about deep learning and structured data?

A. Deep learning cannot process structured data
B. Deep learning is always the best choice for structured data
C. Traditional ML is often sufficient for structured data
D. Structured data requires neural networks

Correct Answer: C

Explanation:
For many structured data problems, traditional machine learning techniques may be simpler and more efficient than deep learning.

---

Question 10

A model uses an input layer, multiple hidden layers, and an output layer. What type of technique does this describe?

A. Clustering
B. Regression
C. Deep learning
D. Rule-based inference

Correct Answer: C

Explanation:
This layered structure is characteristic of deep neural networks, which form the foundation of deep learning techniques.

---

Exam Tips for This Topic

• Look for keywords like images, speech, text, neural networks, and automatic feature extraction
• Avoid choosing deep learning for simple, structured, low-data scenarios
• Remember: deep learning ≠ better in all cases

---

Go to the AI-900 Exam Prep Hub main page.