Free Microsoft Azure AI Fundamentals General Questions and Answers — Questions and Answers

Question 1: You're employed by a car dealership. Your employer requests that you give him forecast data. Will the new auto model succeed or fail? The new model sports a sunroof, improved seats, and many engine upgrades. You created a list of information about previous successful models, including details about their features and sales figures. What should you do to ensure the success of the new model during the pre-processing of the data stage?

Feature selection (Correct answer)
Data selection
Data classification
Training set selection

Correct answer: Feature selection

To predict the success of a new car model, it's crucial to identify which characteristics (features like sunroof, seats, engine upgrades) from historical data are most influential. Feature selection is the process of choosing the most relevant features that contribute significantly to the prediction outcome. By focusing on these key features during data preprocessing, the model's accuracy and efficiency in forecasting success can be greatly improved.

Question 2: An extensive dataset of brain scan images that have been anonymised and classified into different forms of brain hemorrhages is being used in a medical research study. Before the photos are analyzed by a person, machine learning must be used to help early detection of the many forms of brain hemorrhages in the images. Which kind of machine learning does this represent?

classification (Correct answer)
clustering
regression
None of the above

Correct answer: classification

The scenario involves using a dataset of brain scan images that are already 'classified into different forms of brain hemorrhages' to help 'early detection' in new images. This means the machine learning model needs to learn from labeled examples to assign new, unseen images to one of these predefined categories (types of hemorrhages). This task of assigning data points to discrete, predefined classes is known as classification.

Question 3: Why should you randomly divide the rows into several subsets when training a model?

to test the model by using data that was not used to train the model (Correct answer)
to train the model twice to attain better accuracy
to train multiple models simultaneously to attain better performance
All of the above

Correct answer: to test the model by using data that was not used to train the model

Randomly dividing data into separate training and testing (or validation) sets is a fundamental practice in machine learning. The training set is used to build and optimize the model, while the test set, which the model has never encountered, is used to evaluate its real-world performance and generalization ability. This process helps ensure the model can accurately predict on new data and prevents overfitting, where a model performs well only on the data it was trained on.

Question 4: For the following ten years, you must forecast the sea level in meters. Which kind of machine learning ought to be applied?

regression (Correct answer)
clustering
classification
None of these

Correct answer: regression

Forecasting the sea level in meters involves predicting a continuous numerical value over time. Regression is the specific type of machine learning task designed for predicting continuous outcomes. Classification, in contrast, predicts discrete categories, while clustering groups similar data points without predefined labels. Therefore, regression is the appropriate machine learning approach for this prediction problem.

Question 5: What popular ML algorithm do you use when building a Clustering Model?

K-means (Correct answer)
Linear Regression
Two-Class Neural Network
Multiclass Logistic Regression

Correct answer: K-means

K-means is one of the most popular and widely used unsupervised machine learning algorithms for clustering. It works by partitioning 'n' data points into 'k' clusters, where each data point belongs to the cluster with the nearest mean (centroid). The other options, Linear Regression, Two-Class Neural Network, and Multiclass Logistic Regression, are supervised learning algorithms used for regression or classification tasks.

Question 6: What kind of methodology is the Face Recognition system built on?

Strong AI approach
Cognitive AI approach
Applied AI approach (Correct answer)
Weak AI approach

Correct answer: Applied AI approach

Face Recognition systems are practical applications of artificial intelligence designed to solve a specific, real-world problem: identifying or verifying individuals. This focus on developing AI technologies for concrete use cases falls under the 'Applied AI' approach. It contrasts with theoretical AI research (Strong AI) or general cognitive modeling, emphasizing the deployment of AI solutions to address specific challenges.

Question 7: Which of the following describes how artificial intelligence is advantageous?

Helps in providing security
Have the ability to think hence makes the work easier
Reduces the time taken to solve the problem
All of the above (Correct answer)

Correct answer: All of the above

Artificial intelligence offers a wide range of benefits across various sectors. It significantly enhances security through advanced monitoring and threat detection, automates repetitive and complex tasks to improve efficiency and reduce human error, and drastically cuts down the time required to process information and solve problems. Therefore, all the listed points—providing security, making work easier, and reducing problem-solving time—are valid advantages of AI.

Question 8: Which of the following can help an AI agent perform better?

Observing
Learning (Correct answer)
Perceiving
All of the above

Correct answer: Learning

An AI agent's ability to perform better over time is primarily driven by its capacity for learning. Through learning, an agent can adapt its strategies, improve its decision-making, and refine its understanding of the environment based on new data and experiences. While observing and perceiving are essential for gathering information, learning is the process that enables the agent to internalize this information and enhance its overall performance.

Question 9: What platform offers a Conversation AI agent user interface?

QnA Maker
Azure Bot Service (Correct answer)
Computer Vision Service
Bot Framework

Correct answer: Azure Bot Service

The Azure Bot Service provides a comprehensive platform for developing, connecting, and managing intelligent conversational AI agents. It offers a user interface and tools that enable developers to build bots that can interact with users across various channels. While QnA Maker creates knowledge bases and Bot Framework is the underlying SDK, Azure Bot Service is the specific platform that hosts and manages the conversational AI agent's user interface and deployment.

Question 10: Choose every model that makes up Supervised ML.

Classification Model
Regression model
Clustering Model
Both a and b (Correct answer)

Correct answer: Both a and b

Supervised Machine Learning involves training models on labeled datasets, where the input data is paired with the correct output. The two main categories of problems addressed by supervised learning are Classification, which predicts discrete categories (e.g., identifying spam email), and Regression, which predicts continuous numerical values (e.g., forecasting house prices). Clustering, conversely, is an unsupervised learning technique.

Question 11: You are analyzing the line "After Peter met Sara at Microsoft's Paris office, they toured the Eiffel tower" using the Text Analytics Entity Recognition API. What proportion of entities under the category "Location" can you anticipate in the API response?

3 (Correct answer)
2
1
0

Correct answer: 3

Analyzing the sentence 'After Peter met Sara at Microsoft's Paris office, they toured the Eiffel tower' for 'Location' entities reveals three distinct locations. These are 'Microsoft's Paris office' (a specific workplace location), 'Paris' (a city), and 'Eiffel tower' (a famous landmark). The Text Analytics Entity Recognition API would identify each of these as entities falling under the 'Location' category.

Question 12: Your company's customer care department spends a lot of time responding to the same queries. They requested your assistance in automating this process. You received a Microsoft Excel (*.xlsx) document from them that had common queries and usual responses. What service will you employ to turn this document into a knowledge base?

LUIS
QnA Maker (Correct answer)
Text Analytics
Azure Bot Service

Correct answer: QnA Maker

QnA Maker is an Azure AI service specifically designed to create a conversational question-and-answer layer from existing content. It allows you to easily build a knowledge base by importing semi-structured data, such as an Excel document containing common queries and their corresponding answers. This knowledge base can then be integrated into a bot or application to automate responses to frequently asked questions, significantly assisting customer care departments.

Question 13: For the production environment, you are constructing a compute cluster. You've limited the number of nodes to five. What number of nodes should be the bare minimum?

3
1
2
0 (Correct answer)

Correct answer: 0

When configuring a compute cluster in Azure Machine Learning for a production environment, it's possible to set the minimum number of nodes to zero. This 'scale to zero' capability is a cost-saving feature, meaning you only incur charges for compute resources when the cluster is actively processing workloads. When idle, the cluster scales down completely, minimizing operational costs.

Question 14: What markup language will you employ to manage the output of your phone's speech synthesizer?

JSON
SSML (Correct answer)
SQL
HTML

Correct answer: SSML

SSML (Speech Synthesis Markup Language) is an XML-based markup language used to control various aspects of speech synthesis, such as pronunciation, volume, pitch, and speaking rate. When working with speech synthesizers, including those in phones or cloud services like Azure Speech, SSML allows developers to fine-tune the spoken output to create more natural, expressive, and understandable responses.

Question 15: You must use the Azure portal to create a Web Bot. You must select a bot template. Which two SDK dialects are available for the bot template?

C# (Correct answer)
Node.js (Correct answer)
C++
Python

Correct answer: C#

When creating a Web Bot in the Azure portal, the primary SDK dialects offered for bot templates are C# and Node.js. These are the two main programming languages supported by the Azure Bot Framework for developing conversational AI applications. Developers can choose either language to build their bots, leveraging the respective SDKs and tools.

Question 16: What forms the core of the personal digital assistant?

Text Analytics
Azure Speech
Bot Framework (Correct answer)
Computer Vision Service

Correct answer: Bot Framework

The Microsoft Bot Framework serves as the core technology stack for building, connecting, and managing intelligent bots, including personal digital assistants. It provides a comprehensive SDK, tools, and services that enable developers to create conversational AI agents capable of understanding natural language, managing dialogue, and integrating with various platforms and services. It is the foundational platform for bot development.

Free Microsoft Azure AI Fundamentals General Questions and Answers

You're employed by a car dealership.
Your employer requests that you give him forecast data.
Will the new auto model succeed or fail? The new model sports a sunroof, improved seats, and many engine upgrades.
You created a list of information about previous successful models, including details about their features and sales figures.
What should you do to ensure the success of the new model during the pre-processing of the data stage?