Microsoft Azure AI Fundamentals Machine Learning 3 — Questions and Answers
Question 1: Which Azure ML compute target is best suited for running large-scale, distributed training jobs?
- Azure ML Compute Instance
- Azure ML Compute Cluster (Correct answer)
- Azure Container Instance
- Local computer
Correct answer: Azure ML Compute Cluster
Azure ML Compute Clusters scale out automatically to multiple nodes, making them ideal for distributed, large-scale training workloads.
Question 2: What does 'feature engineering' mean in the context of machine learning?
- Choosing the correct ML algorithm
- Transforming or creating input variables to improve model performance (Correct answer)
- Deploying models to production
- Splitting data into training and test sets
Correct answer: Transforming or creating input variables to improve model performance
Feature engineering involves creating, transforming, or selecting input variables (features) from raw data to help a model learn patterns more effectively.
Question 3: In Azure AutoML, which task type should you select to predict a continuous numeric value such as house price?
- Classification
- Clustering
- Regression (Correct answer)
- Anomaly Detection
Correct answer: Regression
Regression predicts continuous numeric outputs, making it the correct AutoML task type for scenarios like predicting house prices or temperatures.
Question 4: What is the purpose of a validation dataset during model training?
- To provide the final unbiased evaluation of the trained model
- To tune hyperparameters and monitor model performance during training (Correct answer)
- To store raw data before preprocessing
- To version-control model artifacts
Correct answer: To tune hyperparameters and monitor model performance during training
A validation dataset is used during training to tune hyperparameters and detect overfitting, while the test set provides the final unbiased evaluation.
Question 5: Which ML concept describes using a model trained on one task as a starting point for a different but related task?
- Ensemble learning
- Transfer learning (Correct answer)
- Reinforcement learning
- Semi-supervised learning
Correct answer: Transfer learning
Transfer learning reuses knowledge from a pretrained model (e.g., ImageNet-trained CNN) and fine-tunes it for a new related task with less data.
Question 6: In Azure ML Designer, what is a 'pipeline' composed of?
- Only data ingestion components
- Modules connected in a graph that represent data processing and model training steps (Correct answer)
- A sequence of REST API calls
- A set of deployment endpoints
Correct answer: Modules connected in a graph that represent data processing and model training steps
Azure ML Designer pipelines are visual graphs made up of connected modules (components) representing steps like data prep, training, and evaluation.
Question 7: What does AUC (Area Under the Curve) measure in model evaluation?
- Training speed of the model
- The model's ability to distinguish between classes across all thresholds (Correct answer)
- The total number of training epochs
- Memory usage during inference
Correct answer: The model's ability to distinguish between classes across all thresholds
AUC measures the area under the ROC curve, summarizing the model's ability to discriminate between positive and negative classes across all classification thresholds.
Microsoft Azure AI Fundamentals Machine Learning 3
Which Azure ML compute target is best suited for running large-scale, distributed training jobs?