Purpose

I created this repository to help you prepare for the AI-900 exam. Unlike other exams, this one is rather simple as long as you understand the content covered in the official learning paths. Working through practice questions helped me pass this exam on the first try. Verify my certification

Official Website: Microsoft Azure AI Fundamentals

Certification (after passing this exam): Microsoft Certified: Azure AI Fundamentals

Practice Questions

Disclaimer: These practice questions are very similar to the actual exam questions in style and skill level; yet are only indicative and by no means a comprehensive list of questions. Questions have not been transcribed from the real exam, which is against exam policy.

1. In this computer vision task, individual pixels in an image are classified according to the object to which they belong.

1. Object Detection
2. Semantic Segmentation
3. Image Analysis
4. Optical Character Recognition

Show Answer

Semantic Segmentation

2. In this computer vision technique, text can be detected and read in images.

1. Object Detection
2. Semantic Segmentation
3. Image Analysis
4. Optical Character Recognition

Show Answer

Optical Character Recognition

3. Which of these can be used to create conversational AI?

1. Q&A Maker
2. Semantic Segmentation
3. Azure Speech Maker
4. Azure Chat Functions

Show Answer

Q&A Maker

4. Which is not one of the resources that can be created using Azure Machine Learning Studio?

1. Compute Instances
2. Compute Balancers
3. Compute Clusters
4. Inference Clusters

Show Answer

Compute Balancers

5. Which would be used to predict the category an item belongs to?

1. Classification
2. K-means
3. Clustering
4. Neural Network

Show Answer

Classification

6. Which technique is used to group similar entities based on features?

1. Classification
2. K-means
3. Clustering
4. Neural Network

Show Answer

Clustering

7. What are the resource types available as Computer Vision service?

1. Computer Vision and Computer Cluster
2. Computer Key and Computer Endpoint
3. Computer Vision and Computer Services
4. Computer Insights and Computer Services

Show Answer

Computer Vision and Computer Services

8. What are the two specialised domain models of Computer Vision?

1. Stars and Places
2. Marker and People
3. Background and Landmarks
4. Celebrities and Landmarks

Show Answer

Celebrities and Landmarks

9. Which of these metrics is not used when model training in Custom Vision?

1. Precision
2. Recall
3. Mean Average Precision
4. F1 Score

Show Answer

F1 Score

10. Which is not a supported image format in the Face Service?

1. JPEG
2. AI
3. PNG
4. BMP

Show Answer

AI (.ai)

11. The OCR API returns hierachical information consisting of all of the following, except:

1. Regions
2. Areas
3. Lines
4. Words

Show Answer

Areas

12. Translation Text uses which model for translation?

1. Neural Network
2. SVM
3. Cognitive Text
4. Neural Machine Translation

Show Answer

Neural Machine Translation

13. Which is not an entity type of LUIS application intents?

1. List
2. RegEx
3. Filter
4. Machine-Learned

Show Answer

Filter

14. Which is not a machine learning algorithm category?

1. Classification
2. Regression
3. Normalization
4. Clustering

Show Answer

Normalization

15. Which is a multiclass classification algorithm?

1. Decision Forest
2. K-means
3. K-nearest Neighbor
4. Anomaly Detection

Show Answer

Decision Forest

16. This is used to highlight the strongest pattern in a dataset:

1. Reinforcement Learning
2. Principal Component Analysis
3. K-means
4. Classification Algorithm

Show Answer

Principal Component Analysis

17. Which is not a supported file format in Azure ML?

1. Excel
2. CSV
3. TSV
4. XML

Show Answer

XML

18. What is the goal of feature engineering?

1. Update values to use a common scale
2. Create a smaller set of discrete ranges
3. Update missing values with other values
4. Match data against the question

Show Answer

Match data against the question

19. The fraction of time when the model is correct is known as:

1. Precision
2. Accuracy
3. Recall
4. F1 Score

Show Answer

Accuracy

20. Which of these confirms how often the model is correct:

1. Precision
2. Accuracy
3. Recall
4. F1 Score

Show Answer

Precision

21 Which value identifies how much the model finds all there is to find?

1. Precision
2. Accuracy
3. Recall
4. F1 Score

Show Answer

Recall

22. This regression type is used to predict a variable that can be considered as a label:

1. Ordinal
2. Linear
3. SVM
4. Poisson

Show Answer

Ordinal

23. This regression type uses counts instead of data values

1. Ordinal
2. Linear
3. SVM
4. Poisson

Show Answer

Poisson

24. Statistical analysis cab be broken down into these three processes:

1. Research, testing, deployment
2. Transformation, Visualisation, Modeling
3. Workflow, Modeling, Deployment
4. ETL, ELT, Testing

Show Answer

Transformation, Visualisation, Modeling

25. Which is not true about hyperparameters?

1. Used to normalize data
2. Defines higher level features of models
3. Are provided as inputs
4. Cannot be directly learned from data using model training processes

Show Answer

Used to normalize data

26. Which is not a metric for clustering models?

1. Mean Absolute Error
2. Average Distance to Cluster Center
3. Average Distance to Other Center
4. Number of Points

Show Answer

Mean Absolute Error

27. Which is not an offering of Cognitive Services?

1. 3D
2. Speech
3. Vision
4. Language

Show Answer

3D

28. Which is a feature of Computer Vision?

1. Image Stabalization
2. Tagging Images
3. Motion Detection
4. Video Thumbnail

Show Answer

Tagging Images

29. According to which principle of responsible AI should AI solutions empower everyone and engage people?

1. Inclusiveness
2. Transparency
3. Accountability
4. Fairness

Show Answer

Inclusiveness