Microsoft Azure AI Fundamentals Certification Exam AI-900 Practice Test 3

Q.

Explanation:

The correct answers are: Statement 1 - Yes, Statement 2 - Yes, Statement 3 - No. Statement 1 is True because Azure Machine Learning designer does provide a drag-and-drop visual interface that allows users to build, test, and deploy machine learning models without writing code by connecting pre-built modules on a visual canvas. Statement 2 is True because the designer enables you to save your machine learning workflows as pipeline drafts, allowing you to preserve your progress, make iterative improvements, and reuse components across different projects. Statement 3 is False because Azure Machine Learning designer does not support custom JavaScript functions - it primarily uses Python-based modules and pre-built components; while you can create custom modules, they would typically be written in Python rather than JavaScript, as the Azure ML platform is primarily Python-based for custom scripting and module development.

Q. A medical research project uses a large anonymized dataset of brain scan images that are categorized into predefined brain haemorrhage types. You need to use machine learning to support early detection of the different brain haemorrhage types in the images before the images are reviewed by a person. This is an example of which type of machine learning?

clustering
regression
classification

Explanation:

The correct answer is C (classification). This scenario describes a supervised learning classification problem where the goal is to categorize brain scan images into predefined brain haemorrhage types. Since the dataset already contains images that are categorized into predefined types, this is a labeled dataset that can be used to train a classification model to predict which category (haemorrhage type) a new brain scan image belongs to. Clustering (A) would be used for unsupervised learning to find hidden patterns or groups in data without predefined categories. Regression (B) is used to predict continuous numerical values, not discrete categories like haemorrhage types.

Q. When training a model, why should you randomly split the rows into separate subsets?

to train the model twice to attain better accuracy
to train multiple models simultaneously to attain better performance
to test the model by using data that was not used to train the model
Check Answer

Explanation:

The correct answer is C (to test the model by using data that was not used to train the model). Randomly splitting data into separate subsets (typically training and testing sets, and sometimes validation sets) is a fundamental practice in machine learning to evaluate model performance objectively. The training set is used to train the model, while the test set is used to evaluate how well the model performs on unseen data. This helps assess the model's ability to generalize to new data and detect overfitting. Option A is incorrect because you don't train the model twice with the same data for better accuracy. Option B is incorrect because splitting data is not about training multiple models simultaneously, but about proper evaluation methodology.

Q. You are evaluating whether to use a basic workspace or an enterprise workspace in Azure Machine Learning. What are two tasks that require an enterprise workspace? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

Use a graphical user interface (GUI) to run automated machine learning experiments.
Create a compute instance to use as a workstation.
Use a graphical user interface (GUI) to define and run machine learning experiments from Azure Machine Learning designer.
Create a dataset from a comma-separated value (CSV) file.
Check Answer

Explanation:

The correct answers are A (Use a graphical user interface (GUI) to run automated machine learning experiments) and C (Use a graphical user interface (GUI) to define and run machine learning experiments from Azure Machine Learning designer). In Azure Machine Learning, enterprise workspaces (now called standard workspaces) include advanced features like the visual designer interface and automated ML GUI capabilities that were not available in basic workspaces. Basic workspaces supported core functionality like creating compute instances (B) and working with datasets from CSV files (D), but lacked the advanced visual interfaces for automated ML and the designer. Note: The workspace tier terminology has evolved since this question was created, but the concept remains that premium features require higher-tier workspace subscriptions.

Q. You need to predict the income range of a given customer by using the following dataset:

Education Level
Last Name
Age
Income Range
First Name
Check Answer

Explanation:

The correct answers are A (Education Level) and C (Age). In machine learning, features are the input variables used to predict the target variable. Since we want to predict Income Range, that field is the target/label, not a feature (eliminating option D). First Name and Last Name (options B and E) are typically not good predictive features for income as individual names don't generally correlate with income patterns. However, Age and Education Level are strong predictive features for income - age can indicate career experience and earning potential, while education level often correlates with income brackets. These two fields contain meaningful information that can help the model learn patterns to predict income ranges for new customers.

Q. You are building a tool that will process images from retail stores and identify the products of competitors. The solution will use a custom model. Which Azure Cognitive Services service should you use?

Custom Vision
Form Recognizer
Face
Computer Vision
Check Answer

Explanation:

The correct answer is A (Custom Vision). Custom Vision is specifically designed for building custom image classification and object detection models. Since the requirement is to identify specific competitor products in retail store images using a custom model, Custom Vision is the ideal service. It allows you to train custom models with your own labeled images to recognize specific products, brands, or objects that are unique to your business needs. Form Recognizer (B) is for extracting text and data from documents and forms, not for product identification. Face (C) is specifically for facial recognition and analysis. Computer Vision (D) provides pre-built image analysis capabilities but doesn't allow you to train custom models for specific product identification - it uses general-purpose models for common objects and scenarios.

Q.

Explanation:

The correct answers are: Statement 1 - Yes, Statement 2 - Yes, Statement 3 - No. Statement 1 is True because organizing documents into groups based on text similarities is a classic example of clustering, an unsupervised learning technique that groups similar items together without predefined categories. Statement 2 is True because grouping patients based on symptoms and diagnostic test results is also clustering, where the algorithm identifies patterns and similarities to group patients with similar medical profiles together. Statement 3 is False because predicting whether a person will develop mild, moderate, or severe allergy symptoms based on pollen count is an example of classification (specifically multi-class classification), not clustering. This involves predicting specific predefined categories (mild/moderate/severe) based on input features (pollen count), which is supervised learning, whereas clustering is unsupervised and discovers hidden groupings without predefined target categories.

Q.

Explanation:

The correct answers are: Statement 1 - No, Statement 2 - Yes, Statement 3 - No. Statement 1 is False because a validation set does not include the input examples used to train the model - it is a separate dataset held back from training and used to evaluate model performance during development and hyperparameter tuning. Statement 2 is True because a validation set is indeed used to determine how well a model predicts labels by providing an unbiased evaluation of the model's performance on unseen data, helping to assess generalization ability and guide model selection. Statement 3 is False because a validation set cannot be used to verify that all training data was used to train the model - the validation set is specifically kept separate from training data to provide an independent assessment; instead, training logs or data tracking mechanisms would be used to verify training data usage.

Q. What are two metrics that you can use to evaluate a regression model? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

coefficient of determination (R2)
F1 score
root mean squared error (RMSE)
area under curve (AUC)
balanced accuracy
Check Answer

Explanation:

The correct answers are A (coefficient of determination (R2)) and C (root mean squared error (RMSE)). These are both standard metrics for evaluating regression models. R2 measures the proportion of variance in the dependent variable that is predictable from the independent variables, with values closer to 1 indicating better model performance. RMSE measures the average magnitude of prediction errors, with lower values indicating better performance. The other options (B, D, E) are classification metrics: F1 score combines precision and recall for classification tasks, AUC (Area Under Curve) measures the performance of binary classification models, and balanced accuracy is used for classification problems with imbalanced datasets.

Q. To complete the sentence, select the appropriate option in the answer area. Predicting how many vehicles will travel across a bridge on a given day is an example of

classification
clustering
regression
Check Answer

Explanation:

The correct answer is C (regression). Predicting how many vehicles will travel across a bridge on a given day involves predicting a continuous numerical value (the count of vehicles), which is a regression problem. Regression is used when the target variable is a continuous numeric value that you want to predict. Classification (A) would be used if you were categorizing something into discrete classes or categories (e.g., high traffic vs. low traffic). Clustering (B) is an unsupervised learning technique used to group similar data points together without predefined labels, which doesn't apply to this prediction scenario.

Q.

Explanation:

The correct sequence for building a model to predict automobile prices in Azure Machine Learning designer is: 1) Automobile price data (Raw) as the data source, 2) Select Columns in Dataset to choose relevant features for prediction, 3) Clean Missing Data to handle any missing values in the dataset, 4) Split Data to divide the dataset into training and testing portions, 5) Linear Regression as the algorithm to train the model for price prediction, 6) Train Model to create the trained model using the training data, and 7) Score Model to evaluate the model's performance on the test data. This sequence follows the standard machine learning workflow: start with raw data, preprocess it by selecting relevant columns and cleaning missing values, split it for training and testing, apply a regression algorithm suitable for predicting continuous values like prices, train the model, and finally score it to assess performance.

Q. Which type of machine learning should you use to identify groups of people who have similar purchasing habits?

classification
regression
clustering
Check Answer

Explanation:

The correct answer is C (clustering). Clustering is an unsupervised learning technique used to identify groups or segments within data based on similarities. When you want to identify groups of people with similar purchasing habits, you're looking to discover natural groupings in the data without predefined categories. Clustering algorithms like K-means, hierarchical clustering, or DBSCAN can analyze purchasing patterns and group customers with similar behaviors together. Classification (A) would be used if you already had predefined categories and wanted to assign new customers to those categories. Regression (B) is used to predict continuous numerical values, not to group data points.

Q. To complete the sentence, select the appropriate option in the answer area. _______ models can be used to predict the sale price of auctioned items.

Classification
Clustering
Regression
Check Answer

Explanation:

The correct answer is C (Regression). Predicting the sale price of auctioned items involves predicting a continuous numerical value (the price), which is a regression problem. Regression models are designed to predict continuous numeric outcomes based on input features such as item characteristics, historical auction data, market conditions, etc. Classification (A) would be used if you were categorizing items into discrete price ranges or categories rather than predicting exact prices. Clustering (B) is an unsupervised learning technique used to group similar items together, not to predict specific values like prices.

Q. Which metric can you use to evaluate a classification model?

True positive rate
mean absolute error (MAE)
coefficient of determination (R2)
root mean squared error (RMSE)
Check Answer

Explanation:

The correct answer is A (True positive rate). True positive rate (also known as sensitivity or recall) is a standard metric for evaluating classification models. It measures the proportion of actual positive cases that were correctly identified by the model. The other options (B, C, D) are all regression metrics: Mean Absolute Error (MAE) measures the average absolute differences between predicted and actual values, coefficient of determination (R2) measures the proportion of variance explained by the model, and Root Mean Squared Error (RMSE) measures the average magnitude of prediction errors. These regression metrics are not appropriate for evaluating classification models, which deal with discrete categories rather than continuous values.

Q. Which two components can you drag onto a canvas in Azure Machine Learning designer? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

dataset
compute
pipeline
module
Check Answer

Explanation:

The correct answers are A (dataset) and D (module). In Azure Machine Learning designer, you can drag datasets and modules onto the visual canvas to build machine learning pipelines. Datasets represent the data sources that you want to use in your pipeline, while modules are the functional components that perform specific operations like data transformation, model training, or evaluation. You connect these components together to create a complete machine learning workflow. Compute (B) refers to the computational resources used to run the pipeline, but it's not a component you drag onto the canvas. Pipeline (C) is the entire workflow itself, not a component that you drag onto it.

Q. You need to create a training dataset and validation dataset from an existing dataset. Which module in the Azure Machine Learning designer should you use?

Select Columns in Dataset
Add Rows
Split Data
Join Data
Check Answer

Explanation:

The correct answer is C (Split Data). The Split Data module in Azure Machine Learning designer is specifically designed to divide a single dataset into multiple subsets, typically for creating training and validation datasets. You can specify the splitting ratio (e.g., 80% for training, 20% for validation) and choose whether to split randomly or based on specific criteria. Select Columns in Dataset (A) is used to choose specific columns from a dataset, not to split rows. Add Rows (B) is used to combine datasets by adding rows from one dataset to another. Join Data (D) is used to combine datasets based on matching key columns, similar to SQL joins.

Q.

Explanation:

The correct matches are: 'Predict how many minutes late a flight will arrive based on the amount of snowfall at an airport' should be matched with Regression because it involves predicting a continuous numerical value (minutes of delay). 'Segment customers into different groups to support a marketing department' should be matched with Clustering because it involves grouping similar customers together without predefined categories, which is an unsupervised learning technique. 'Predict whether a student will complete a university course' should be matched with Classification because it involves predicting a categorical outcome (complete or not complete), which is a binary classification problem where the model assigns data points to predefined categories.

Q. To complete the sentence, select the appropriate option in the answer area. Hot Area: _______ is the calculated probability of a correct image classification

Accuracy
Confidence
Root Mean Square Error
Sentiment
Check Answer

Explanation:

The correct answer is A (Accuracy). Accuracy is indeed the calculated probability of a correct image classification, representing the ratio of correctly classified images to the total number of images. Confidence (B) refers to how certain the model is about its prediction, not the probability of correctness. Root Mean Square Error (C) is a metric used for regression problems, not classification. Sentiment (D) is a type of classification task, not a metric for measuring classification performance.

Q. To complete the sentence, select the appropriate option in the answer area. Hot Area: Ensuring an AI system does not provide a prediction when important fields contain unusual or missing values is _______ principle for responsible AI.

an inclusiveness
a privacy and security
a reliability and safety
a transparency
Check Answer

Explanation:

The correct answer is C (a reliability and safety). Ensuring an AI system does not provide predictions when important fields contain unusual or missing values is a key aspect of reliability and safety in responsible AI. This practice prevents the system from making potentially harmful or inaccurate predictions based on incomplete or anomalous data, thereby maintaining system reliability and user safety. Inclusiveness (A) focuses on ensuring AI systems work for all users regardless of their abilities or backgrounds. Privacy and security (B) deals with protecting user data and system security. Transparency (D) involves making AI systems understandable and explainable to users.

Q. To complete the sentence, select the appropriate option in the answer area. Hot Area: Ensuring that the numeric variables in training data are on a similar scale is an example of _______.

data ingestion
feature engineering
feature selection
model training
Check Answer

Explanation:

The correct answer is B (feature engineering). Ensuring that numeric variables are on a similar scale refers to data normalization or standardization techniques (like min-max scaling or z-score normalization), which are key components of feature engineering. This process transforms the existing features to improve model performance by preventing features with larger scales from dominating those with smaller scales. Data ingestion (A) refers to the process of collecting and importing data. Feature selection (C) involves choosing the most relevant features from the dataset. Model training (D) is the process of teaching the algorithm using the prepared data.

Q. To complete the sentence, select the appropriate option in the answer area. Hot Area: Assigning classes to images before training a classification model is an example of _______.

evaluation
feature engineering
hyperparameter tuning
labeling
Check Answer

Explanation:

The correct answer is D (labeling). Assigning classes to images before training a classification model is the process of data labeling or annotation, where each image is tagged with its correct class or category. This creates the ground truth that supervised learning algorithms need to learn from during training. Evaluation (A) refers to assessing model performance after training. Feature engineering (B) involves transforming or creating new features from existing data. Hyperparameter tuning (C) is the process of optimizing model parameters to improve performance.

Q. Your company wants to build a recycling machine for bottles. The recycling machine must automatically identify bottles of the correct shape and reject all other items. Which type of AI workload should the company use?

anomaly detection
conversational AI
computer vision
natural language processing
Check Answer

Explanation:

The correct answer is C (computer vision). Computer vision is the AI workload specifically designed to analyze and interpret visual data from images or video streams. For a recycling machine that needs to automatically identify bottles of the correct shape and reject other items, computer vision techniques like image classification, object detection, and shape recognition would be essential. Anomaly detection (A) identifies unusual patterns in data but wouldn't be the primary approach for shape identification. Conversational AI (B) deals with human-computer dialogue through speech or text. Natural language processing (D) focuses on understanding and processing human language, not visual recognition tasks.

Q.

Explanation:

The correct matches are: 'Predict how many minutes late a flight will arrive based on the amount of snowfall at an airport' should be matched with Regression because it involves predicting a continuous numerical value (minutes of delay). 'Segment customers into different groups to support a marketing department' should be matched with Clustering because it involves grouping similar customers together without predefined categories, which is an unsupervised learning technique. 'Predict whether a student will complete a university course' should be matched with Classification because it involves predicting a categorical outcome (complete or not complete), which is a binary classification problem where the model assigns data points to predefined categories.

Q. In which two scenarios can you use the Form Recognizer service? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

Extract the invoice number from an invoice.
Translate a form from French to English.
Find image of product in a catalog.
Identify the retailer from a receipt.
Check Answer

Explanation:

The correct answers are A (Extract the invoice number from an invoice) and D (Identify the retailer from a receipt). Form Recognizer is an Azure AI service specifically designed to extract text, key-value pairs, and structured data from documents and forms. It can analyze invoices to extract specific information like invoice numbers, dates, and amounts, and it can process receipts to identify merchant information, item details, and totals. Option B (Translate a form from French to English) would require Azure Translator service, not Form Recognizer. Option C (Find image of product in a catalog) would require computer vision or visual search capabilities, not document analysis.

Q.

Explanation:

The correct matches are: 'Predict how many minutes late a flight will arrive based on the amount of snowfall at an airport' should be matched with Regression because it involves predicting a continuous numerical value (minutes of delay). 'Segment customers into different groups to support a marketing department' should be matched with Clustering because it involves grouping similar customers together without predefined categories, which is an unsupervised learning technique. 'Predict whether a student will complete a university course' should be matched with Classification because it involves predicting a categorical outcome (complete or not complete), which is a binary classification problem where the model assigns data points to predefined categories.

Q. You need to develop a mobile app for employees to scan and store their expenses while traveling. Which type of computer vision should you use?

semantic segmentation
image classification
object detection
optical character recognition (OCR)
Check Answer

Explanation:

The correct answer is D (optical character recognition (OCR)). For a mobile app that needs to scan and store expense information from receipts, invoices, or other documents, OCR is the most appropriate computer vision technique. OCR extracts text data from images, allowing the app to read and digitize information like merchant names, dates, amounts, and item descriptions from expense documents. Semantic segmentation (A) divides images into meaningful regions but doesn't extract text. Image classification (B) categorizes entire images but doesn't extract specific data. Object detection (C) identifies and locates objects in images but doesn't read text content.

Q.

Explanation:

The correct answers are: Statement 1 - Yes, Statement 2 - Yes, Statement 3 - No. Statement 1 is True because the Custom Vision service is specifically designed for object detection and image classification tasks, allowing you to detect and identify objects within images by training custom models with your own labeled data. Statement 2 is True because Custom Vision requires you to provide your own training data - you must upload and label images that represent the objects or categories you want the model to recognize, as it doesn't come with pre-trained models for specific custom use cases. Statement 3 is False because Custom Vision is designed for analyzing static images, not video files; while you could extract individual frames from videos to analyze as separate images, the service itself does not directly process or analyze video files as a continuous stream.

Q. You are processing photos of runners in a race. You need to read the numbers on the runners' shirts to identify the runners in the photos. Which type of computer vision should you use?

facial recognition
optical character recognition (OCR)
semantic segmentation
object detection
Check Answer

Explanation:

The correct answer is B (optical character recognition (OCR)). Since the task specifically requires reading the numbers on runners' shirts to identify them, OCR is the appropriate computer vision technique. OCR is designed to extract and recognize text and numeric characters from images, making it perfect for reading race numbers printed on athletic bibs or shirts. Facial recognition (A) would identify people by their faces, not by reading numbers. Semantic segmentation (C) divides images into meaningful regions but doesn't read text. Object detection (D) can locate objects like runners but doesn't extract the numeric text from their shirts.

Q.

Explanation:

The correct answers are: 'Image classification' for separating images of polar bears and brown bears, 'Object detection' for determining the location of a bear in a photo, and 'Semantic segmentation' for determining which pixels in an image are part of a bear. 'Image classification' is used to categorize entire images into predefined classes based on the overall content, which makes it suitable for distinguishing between types of bears. 'Object detection' identifies and locates objects within an image by placing bounding boxes around them, making it ideal for spotting the location of a bear. 'Semantic segmentation' provides pixel-level classification, assigning a label to each pixel in the image, which is necessary for tasks where the exact shape and area of an object €”like a bear €”must be delineated.

Q. You use drones to identify where weeds grow between rows of crops to send an instruction for the removal of the weeds. This is an example of which type of computer vision?

object detection
optical character recognition (OCR)
scene segmentation
Check Answer

Explanation:

The correct answer is C (scene segmentation). Scene segmentation (also known as semantic segmentation) is the computer vision technique that divides an image into meaningful regions or segments, classifying each pixel according to what it represents. In this agricultural scenario, the system needs to segment the aerial images into different categories like crops, weeds, soil, and pathways between rows. This pixel-level classification allows precise identification of where weeds are growing relative to the crop rows. Object detection (A) would identify discrete objects but wouldn't provide the detailed spatial mapping needed for precise weed location. OCR (B) is for reading text and numbers, which is not relevant to this agricultural application.

Q.

Explanation:

The correct answers are: 'Verification' for 'Do two images of a face belong to the same person?', 'Similarity' for 'Does this person look like other people?', 'Grouping' for 'Do all the faces belong together?', and 'Identification' for 'Who is this person in this group of people?'. 'Verification' is used to compare two face images and determine if they belong to the same individual. 'Similarity' involves finding faces that resemble a given face, often used in search or matching scenarios. 'Grouping' clusters similar faces together without prior labeling, which helps determine if faces belong to the same group. 'Identification' attempts to determine the identity of a face by comparing it to a known set of faces, answering the question of 'who' the person is within a group.

Q.

Explanation:

The correct answers are: 'Facial recognition' for identifying celebrities in images, 'Optical character recognition (OCR)' for extracting movie title names from poster images, and 'Object detection' for locating vehicles in images. 'Facial recognition' is a computer vision task that matches detected faces to known identities, making it suitable for identifying celebrities. 'OCR' is used to detect and extract text from images, such as movie titles on posters. 'Object detection' identifies and localizes objects, such as vehicles, within an image using bounding boxes. Each workload type is matched with its specific function in the scenario.

Q. You need to determine the location of cars in an image so that you can estimate the distance between the cars. Which type of computer vision should you use?

optical character recognition (OCR)
object detection
image classification
face detection
Check Answer

Explanation:

The correct answer is B (object detection). Object detection is the computer vision technique that identifies and locates specific objects (in this case, cars) within an image by drawing bounding boxes around them and providing their coordinates. This spatial information is essential for determining the precise locations of cars and calculating distances between them. OCR (A) is for reading text and characters, not for locating objects. Image classification (C) categorizes entire images but doesn't provide location information about individual objects within the image. Face detection (D) is specifically for detecting human faces, not cars.

Q.

Explanation:

The correct answer is B (Custom Vision). Azure Custom Vision is a cognitive service that enables you to build, deploy, and improve your own image classifiers by training them with your own images. It allows you to label images and train a machine learning model to detect objects based on those labels. Unlike the Computer Vision service, which only provides prebuilt features and analysis (like identifying objects, brands, or adult content), Custom Vision allows for custom training, testing, and retraining based on specific project needs. Form Recognizer (C) is used for extracting structured data from documents like forms or invoices, and Video Indexer (D) is designed to analyze video content, not for training object detection models.

Q. You send an image to a Computer Vision API and receive back the annotated image shown in the exhibit. Which type of computer vision was used?

object detection
semantic segmentation
optical character recognition (OCR)
image classification
Check Answer

Explanation:

The correct answer is A (object detection). The annotated image clearly shows the characteristics of object detection: multiple objects (banana, orange, apple) are identified with bounding boxes drawn around each object, along with class labels and confidence scores (97.56%, 95.77%, 98.21%). Object detection not only classifies what objects are present but also localizes where they are in the image using rectangular bounding boxes. Semantic segmentation (B) would show pixel-level masks rather than bounding boxes. OCR (C) is for reading text, not identifying fruits. Image classification (D) would only provide overall image labels without localization information.

Q. What are two tasks that can be performed by using the Computer Vision service? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

Train a custom image classification model.
Detect faces in an image.
Recognize handwritten text.
Translate the text in an image between languages.
Check Answer

Explanation:

The correct answers are B (Detect faces in an image) and C (Recognize handwritten text). Azure Computer Vision service provides pre-built capabilities including face detection, which can identify and locate human faces in images, and OCR (Optical Character Recognition) that can extract both printed and handwritten text from images. Option A (Train a custom image classification model) would require Azure Custom Vision service, not the standard Computer Vision service. Option D (Translate text between languages) would require Azure Translator service after the text has been extracted; Computer Vision can read the text but cannot translate it.

Q. What is a use case for classification?

predicting how many cups of coffee a person will drink based on how many hours the person slept the previous night.
analyzing the contents of images and grouping images that have similar colors
predicting whether someone uses a bicycle to travel to work based on the distance from home to work.
predicting how many minutes it will take someone to run a race based on past race times
Check Answer

Explanation:

The correct answers are B (analyzing the contents of images and grouping images that have similar colors) and C (predicting whether someone uses a bicycle to travel to work based on the distance from home to work). Classification algorithms predict discrete categories or classes. Option B involves categorizing images into groups based on color similarity, and Option C predicts a binary outcome (yes/no for bicycle usage). Options A and D are regression problems because they predict continuous numerical values - the number of cups of coffee and the number of minutes, respectively. Classification deals with categorical outcomes, while regression deals with continuous numerical predictions.

Q. What are two tasks that can be performed by using computer vision? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

Predict stock prices.
Detect brands in an image.
Detect the color scheme in an image
Translate text between languages.
Extract key phrases.
Check Answer

Explanation:

The correct answers are B (Detect brands in an image) and C (Detect the color scheme in an image). Computer vision is specifically designed to analyze and interpret visual content. Brand detection involves object recognition and logo identification within images, while color scheme detection involves analyzing the visual properties and color distribution of images. Option A (Predict stock prices) is a time series forecasting problem, not a computer vision task. Option D (Translate text between languages) is a natural language processing task that requires translation services. Option E (Extract key phrases) is also an NLP task for text analysis, not visual analysis.

Q. You need to build an image tagging solution for social media that tags images of your friends automatically. Which Azure Cognitive Services service should you use?

Face
Form Recognizer
Text Analytics
Computer Vision
Check Answer

Explanation:

The correct answer is A (Face). Azure Face service is specifically designed for facial recognition and identification tasks. It can detect, recognize, and identify people in images, making it the ideal choice for automatically tagging friends in social media photos. The Face service can learn to recognize specific individuals and then identify them in new images. Form Recognizer (B) is for extracting data from documents and forms. Text Analytics (C) processes and analyzes text, not images. While Computer Vision (D) can detect faces, it doesn't provide the facial recognition and identification capabilities needed to tag specific friends automatically.

Q. In which two scenarios can you use the Form Recognizer service? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.

Identify the retailer from a receipt
Translate from French to English
Extract the invoice number from an invoice
Find images of products in a catalog
Check Answer

Explanation:

The correct answers are A (Identify the retailer from a receipt) and C (Extract the invoice number from an invoice). Azure Form Recognizer is specifically designed to extract text, key-value pairs, and structured data from documents and forms. It can analyze receipts to identify merchant information, dates, totals, and other structured data, and it can process invoices to extract specific fields like invoice numbers, vendor information, and line items. Option B (Translate from French to English) would require Azure Translator service, not Form Recognizer. Option D (Find images of products in a catalog) would require computer vision or visual search capabilities, not document analysis functionality.