AI-900 practice exam

classification NOTES: Classification is used to predict categories of data. It can predict which category or class an item of data belongs to. In this example, sentiment analysis can be carried out on the Twitter posts with a numeric value applied to the posts to identify and classify positive or negative sentiment. Clustering is a machine learning type that analyzes unlabeled data to find similarities in the data. Regression is a machine learning scenario that is used to predict numeric values. Data transformation is not a machine learning type.

A company deploys an online marketing campaign to social media platforms for a new product launch. The company wants to use machine learning to measure the sentiment of users on the Twitter platform who made posts in response to the campaign. Which type of machine learning is this? Select only one answer. classification clustering data transformation regression

reliability and safety NOTES: The reliability and safety principles are of paramount importance here as it requires an AI system to work alongside people in a physical environment by using AI controlled machinery. The system must function safely, while ensuring no harm will come to human life.

A company is currently developing driverless agriculture vehicles to help harvest crops. The vehicles will be deployed alongside people working in the crop fields, and as such, the company will need to carry out robust testing. Which principle of responsible artificial intelligence (AI) is most important in this case? Select only one answer. accountability Inclusiveness reliability and safety transparency

weather temperature weekday or weekend NOTES: Weather temperature and weekday or weekend are features that provide a weather temperature for a given day and a value based on whether the day is on a weekend or weekday. These are input variables for the model to help predict the labels for e-scooter battery levels, number of hires, and distance traveled. E-scooter battery levels, number of hires, and distance traveled are numeric labels you are attempting to predict through the machine learning model.

A company is using machine learning to predict various aspects of its e-scooter hire service dependent on weather. This includes predicting the number of hires, the average distance traveled, and the impact on e-scooter battery levels. For the machine learning model, which two attributes are the features? Each correct answer presents a complete solution. Select all answers that apply. distance traveled e-scooter battery levels e-scooter hires weather temperature weekday or weekend

water use

A company wants to predict household water use based on the number of people in a house, the weather temperature, and the time of year. In terms of data labels and features, what is the label in this use case? number of people in the house time of year water use weather temperature

clustering NOTES: Clustering is a machine learning type that analyzes unlabeled data to find similarities present in the data. It then groups (clusters) similar data together. In this example, the company can group online customers based on attributes that include demographic data and shopping behaviors. The company can then recommend new products to those groups of customers who are most likely to be interested in them. Classification and multiclass classification are used to predict categories of data. Regression is a machine learning scenario that is used to predict numeric values.

A retailer wants to group together online shoppers that have similar attributes to enable its marketing team to create targeted marketing campaigns for new product launches. Which type of machine learning is this? classification clustering multiclass classification regression

regression NOTES: Regression is a machine learning scenario that is used to predict numeric values. In this example, regression will be able to predict future energy consumption based on analyzing historical time-series energy data based on factors, such as seasonal weather and holiday periods. Multiclass classification is used to predict categories of data. Clustering analyzes unlabeled data to find similarities present in the data. Classification is used to predict categories of data.

An electricity utility company wants to develop a mobile app for its customers to monitor their energy use and to display their predicted energy use for the next 12 months. The company wants to use machine learning to provide a reasonably accurate prediction of future energy use by using the customers' previous energy-use data. Which type of machine learning is this? Select only one answer. classification clustering multiclass classification regression

Identify potential harms. NOTES: Identifying potential harms is the first stage when planning a responsible generative AI solution.

As per the NIST AI Risk Management Framework, what is the first stage to consider when developing a responsible generative AI solution? Select only one answer. Identify potential harms. Measure the presence of potential harms. Mitigate potential harms. Operate the solution.

Embeddings NOTES: Embeddings is an Azure OpenAI model that converts text into numerical vectors for analysis. Embeddings can be used to search, classify, and compare sources of text for similarity.

Select the answer that correctly completes the sentence. [Answer choice] can search, classify, and compare sources of text for similarity. Select only one answer. Data grounding Embeddings Machine learning System messages

system messages NOTES: System messages should be used to set the context for the model by describing expectations. Based on system messages, the model knows how to respond to prompts. The other techniques are also used in generative AI models, but for other use cases.

Select the answer that correctly completes the sentence. [Answer choice] can used to identify constraints and styles for the responses of a generative AI model. Select only one answer. Data grounding Embeddings System messages Tokenization

Copilots NOTES: Copilots are often integrated into applications to provide a way for users to get help with common tasks from a generative AI model. Copilots are based on a common architecture, so developers can build custom copilots for various business-specific applications and services.

Select the answer that correctly completes the sentence. [Answer choice] use plugins to provide end users with the ability to get help with common tasks from a generative AI model. Select only one answer. Copilots Language Understanding solutions Question answering models RESTful API services

Custom model A pre-build receipt model Overall explanation Form Recognizer service is one of the Azure Computer vision solutions additional to Computer Vision service, Custom Vision Service and Face service. For automated document processing, Form Recognizer uses two models: Custom Model and a pre-build receipt model. With the Custom Model approach, you train the Form Recognizer model based on your own form and data. You just need only 5 samples of your form to start. A pre-build receipt model is a Form Recognizer default model that is trained to work with receipts. It helps recognize receipts and extract data from them.

What are two automated document processing models Form Recognizer supports? Custom model Custom vision Standard cognitive model A pre-build receipt model Auto OCR model

optical character recognition NOTES: -OCR is used to extract text and handwriting from images. In this case, it can be used to extract signatures for attendance purposes. -Face detection can detect and verify human faces, not text, from images. -Object detection can detect multiple objects in an image by using bounding box coordinates. It is not used to extract handwritten text. -Image classification is the part of computer vision that is concerned with the primary contents of an image.

What can be used for an attendance system that can scan handwritten signatures? face detection image classification object detection optical character recognition (OCR)

NaN NOTES: NaN, or not a number, designates an unknown confidence score. Unknown is a value with which the NaN confidence score is associated. The score values range between 0 and 1, with 0 designating the lowest confidence score and 1 designating the highest confidence score.

What is the confidence score returned by the Azure AI Language detection service of natural language processing (NLP) for an unknown language name? Select only one answer. 1 -1 NaN Unknown

removing stop words NOTES: Removing stop words is the first step in the statistical analysis of terms used in a text in the context of NLP. Counting the occurrences of each word takes place after stop words are removed. Creating a vectorized model is not part of statistical analysis. It is used to capture the sematic relationship between words. Encoding words as numeric features is not part of statistical analysis. It is frequently used in sentiment analysis.

What is the first step in the statistical analysis of terms in a text in the context of natural language processing (NLP)? Select only one answer. creating a vectorized model counting the occurrences of each word encoding words as numeric features removing stop words

1.0 Overall explanaltion Area Under Curve (AUC) is the model performance metrics for classification models. For binary classification models, the AUC value of 0.5 represents the random predictions. The model predictions are the same as randomly selected values of "Yes" or "No." If the AUC value is below 0.5, the model performance is worse than random. Ideally, the best-fitted model has a value of 1. Such an ideal model predicts all the values correctly.

What is the ideal value for AUC? 0.5 0 0.1 0.75 2 1.0

Anomaly Detection Overall explanation Customers are using Anomaly detection APIs for constant monitoring of their time-series data. An anomaly detection service ingests the data and automatically selects the best ML model for the identification of the possible data irregularities. The service alerts the customers as soon as such anomalies arise. Anomaly Detection is one of the five key elements of Microsoft Artificial Intelligence. The other four are Machine Learning, Computer Vision, Natural Language Processing, and Conversational AI. Machine Learning, Computer Vision, Natural Language Processing, and Conversational AI, along with Anomaly detection, are the key elements of Artificial Intelligence. Automated Machine Learning is a feature of Machine Learning and is not part of Anomaly detection.

What is the name of the common AI service that provides 24/7 monitoring of the customer's time-series data for possible data irregularities? Automated Machine Learning Machine Learning Conversational AI Computer Vision Natural Language Processing Anomaly Detection

Reliability & Safety Overall explanation Microsoft recognizes six principles of responsible AI: Fairness, Reliability and safety, Privacy and security, Transparency, Inclusiveness and Accountability. The principle of Reliability and safety directs AI solutions to respond safely to non-standard situations and to resist harmful manipulations.

What is the name of the responsible AI principle that directs AI solutions design to include resistance to harmful manipulation? Fairness Reliability & Safety Transparency Accountability Inclusiveness Privacy & security

evaluating the trained model NOTES: The validation dataset is a sample of data held back from a training dataset. It is then used to evaluate the performance of the trained model. Cleaning missing data is used to detect missing values and perform operations to fix the data or create new values. Feature engineering is part of preparing the dataset and related data transformation processes. Summarizing the data is used to provide summary statistics, such as the mean or count of distinct values in a column.

What is the purpose of a validation dataset used for as part of the development of a machine learning model? Select only one answer. cleaning missing data evaluating the trained model Feature engineering summarizing the data

evaluate the trained model NOTES: The validation dataset is a sample of data held back from a training dataset. It is then used to evaluate the performance of the trained model. Cleaning missing data is used to detect missing values and perform operations to fix the data or create new values. Feature engineering is part of preparing the dataset and related data transformation processes. Summarizing the data is used to provide summary statistics, such as the mean or count of distinct values in a column.

What is the purpose of a validation dataset used for as part of the development of a machine learning model? Select only one answer. cleaning missing data evaluating the trained model feature engineering summarizing the data

deep learning NOTES: Modern image classification solutions are based on deep learning techniques. Semantic segmentation provides the ability to classify individual pixels in an image depending on the object that they represent. Both linear regression and multiple linear regression use training and validating predictions to predict numeric values, so they are not part of image classification solutions.

Which artificial intelligence (AI) technique serves as the foundation for modern image classification solutions? Select only one answer. semantic segmentation deep learning linear regression multiple linear regression

c. optical character recognition

Which artificial intelligence (AI) technique should be used to extract the name of a store from a photograph displaying the store front? a. image classification b. natural language processing (NLP) c. optical character recognition (OCR) d. semantic segmentation

optical character recognition (OCR) NOTES: OCR provides the ability to detect and read text in images. NLP is an area of AI that deals with identifying the meaning of a written or spoken language, but not detecting or reading text in images. Image classification classifies images based on their contents. Semantic segmentation provides the ability to classify individual pixels in an image. Understand computer vision - Training | Microsoft Learn

Which artificial intelligence (AI) technique should be used to extract the name of a store from a photograph displaying the store front? Select only one answer. image classification natural language processing (NLP) optical character recognition (OCR) semantic segmentation

extracting key phrases from a business insights report NOTES: Extracting key phrases from text to identify the main terms is an NLP workload. Predicting whether customers are likely to buy a product based on previous purchases requires the development of a machine learning model. Monitoring for sudden increases in quantity of failed sign-in attempts is a different workload. Identifying objects in landscape images is a computer vision workload.

Which artificial intelligence (AI) workload scenario is an example of natural language processing (NLP)? Select only one answer. a. extracting key phrases from a business insights report b. identifying objects in landscape images c. monitoring for sudden increases in quantity of failed sign-in attempts d. predicting whether customers are likely to buy a product based on previous purchases

Features are independent of each other NOTES: Multiple linear regression models the relationship between several features and a single label. The features must be independent of each other, otherwise, the model's predictions will be misleading.

Which assumption of the multiple linear regression model should be satisfied to avoid misleading predictions? Select only one answer. Features are dependent on each other Features are independent of each other. Labels are dependent on each other. Labels are independent of each other.

facial detection NOTES: Facial detection provides the ability to detect and analyze human faces in an image, including identifying a person's age based on a photograph. Image classification classifies images based on their contents. Object detection provides the ability to generate bounding boxes identifying the locations of different types of vehicles in an image. Semantic segmentation provides the ability to classify individual pixels in an image.

Which computer vision solution provides the ability to identify a person's age based on a photograph? Select only one answer. facial detection image classification object detection semantic segmentation

use of historical data with known label values to train a model NOTES: Regression is an example of supervised machine learning due to the use of historical data with known label values to train a model. Regression does not rely on randomly generated data for training.

Which feature makes regression an example of supervised machine learning? Select only one answer. use of historical data with known label values to train a model use of historical data with unknown label values to train a model use of randomly generated data with known label values to train a model use of randomly generated data with unknown label values to train a model

tagging NOTES: - Tagging involves associating an image with metadata that summarizes the attributes of the image. - Detecting image types involves identifying clip art images or line drawings. -- Content organization involves identifying people or objects in photos and organizing them based on the identification. - Categorizing involves associating the contents of an image with a limited set of categories.

Which feature of computer vision involves associating an image with metadata that summarizes the attributes of the image? categorizing content organization detecting image types tagging

entity recognition NOTES: Entity recognition includes the entity linking functionality that returns links to external websites to disambiguate terms (entities) identified in a text. Key phrase extraction evaluates the text of a document and identifies its main talking points. Azure AI Language detection identifies the language in which text is written. Sentiment analysis evaluates text and returns sentiment scores and labels for each sentence.

Which feature of the Azure AI Language service includes functionality that returns links to external websites to disambiguate terms identified in a text? Select only one answer. entity recognition key phrase extraction language detection sentiment analysis

speech recognition NOTES: Speech recognition uses audio data to analyze speech and determine recognizable patterns that can be mapped to distinct user voices. Azure AI Speech synthesis is concerned with vocalizing data, usually by converting text to speech. Azure AI Speech translation is concerned with multilanguage translation of speech. Language identification is used to identify languages spoken in audio when compared against a list of supported languages.

Which feature of the Azure AI Speech service can identify distinct user voices? Select only one answer. language identification speech recognition speech synthesis speech translation

DALL-E NOTES: DALL-E is a model that can generate images from natural language. GPT-4 and GPT-3.5 can understand and generate natural language and code but not images. Embeddings can convert text into numerical vector form to facilitate text similarity. Whisper can transcribe and translate speech to text.

Which generative AI model is used to generate images based on natural language prompts? Select only one answer. DALL-E Embeddings GPT-3.5 GPT-4 Whisper

vectorization NOTES: - Vectorization captures semantic relationships between words by assigning them to locations in n-dimensional space. - Lemmatization, also known as stemming, normalizes words before counting them. - Frequency analysis counts how often a word appears in a text. - N-grams extend frequency analysis to include multi-term phrases.

Which natural language processing (NLP) technique assigns values to words such as plant and flower, so that they are considered closer to each other than a word such as airplane? Select only one answer. frequency analysis lemmatization N-grams vectorization

stemming NOTES: - Stemming normalizes words before counting them. - Frequency analysis counts how often a word appears in a text. - N-grams extend frequency analysis to include multi-term phrases. - Vectorization captures semantic relationships between words by assigning them to locations in n-dimensional space.

Which natural language processing (NLP) technique normalizes words before counting them? frequency analysis N-grams stemming vectorization

tokenization NOTES: - Tokenization is part of speech synthesis that involves breaking text into individual words such that each word can be assigned phonetic sounds. - Transcribing is part of speech recognition, which involves converting speech into a text representation. - Key phrase extraction is part of language processing, not speech synthesis. - Lemmatization, also known as stemming, is part of language processing, not speech synthesis.

Which part of speech synthesis in natural language processing (NLP) involves breaking text into individual words such that each word can be assigned phonetic sounds? Select only one answer. lemmatization key phrase extraction tokenization transcribing

accountability NOTES: The accountability principle ensures that AI systems are designed to meet any ethical and legal standards that are applicable. The privacy and security principle states that AI systems must be designed to protect any personal and/or sensitive data. The inclusiveness principle states that AI systems must empower people in a positive and engaging way. The fairness principle is applied to AI system to ensure that users of the systems are treated fairly.

Which principle of responsible artificial intelligence (AI) ensures that an AI system meets any legal and ethical standards it must abide by? Select only one answer. accountability fairness inclusiveness privacy and security

fairness NOTES: Fairness is meant to ensure that AI models do not unintentionally incorporate a bias based on criteria such as gender or ethnicity. Transparency does not apply in this case since banks commonly use their proprietary models when processing loan approvals. Inclusiveness is also out of scope since not everyone is qualified for a loan. Safety is not a primary consideration since there is no direct threat to human life or health in this case.

Which principle of responsible artificial intelligence (AI) plays the primary role when implementing an AI solution that meet qualifications for business loan approvals? Select only one answer. accountability fairness inclusiveness safety

transparency NOTES: Transparency provides clarity regarding the purpose of AI solutions, the way they work, as well as their limitations. The privacy and security, reliability and safety, and accountability principles focus on the capabilities of AI, rather than raising awareness about its limitations.

Which principle of responsible artificial intelligence (AI) raises awareness about the limitations of AI-based solutions? Select only one answer. accountability privacy and security reliability and safety transparency

Azure AI Custom Vision NOTES: - Azure AI Custom Vision is an image recognition service that allows you to build and deploy your own image models. - The Azure AI vision service, Azure AI Face service, and Azure AI Language service do not provide the capability to train your own image model.

Which service can you use to train an image classification model? Select only one answer. Azure AI Vision Azure AI Custom Vision Azure AI Face Azure AI Language

NOTES: creating variations of an image editing an image new image creation

Which three capabilities are examples of image generation features for a generative AI model? Each correct answer presents a complete solution. Select all answers that apply. animation of static images creating variations of an image editing an image extracting RGB values from an image new image creation

Entity Linking Personally Identifiable Information (PII) detection Sentiment analysis NOTES: Entity Linking, PII detection, and sentiment analysis are all elements of the Azure AI Service for Azure AI Language. Azure AI Vision deals with image processing. Azure AI Content Moderator is an Azure AI Services service that is used to check text, image, and video content for material that is potentially offensive.

Which three features are elements of the Azure AI Language Service? Each correct answer presents a complete solution. Select all answers that apply. Azure AI Vision Azure AI Content Moderator Entity Linking Personally Identifiable Information (PII) detection Sentiment analysis

language identification speaker recognition voice assistants NOTES: Language identification, speaker recognition, and voice assistants are all elements of the Azure AI Speech service. Text translation and document translation are part of the Translator service.

Which three features are elements of the Azure AI Speech service? Each correct answer presents a complete solution. Select all answers that apply. document translation language identification speaker recognition text translation voice assistants

choosing a model, evaluating a model, training a model NOTES: The computer vision service eliminates the need for choosing, training, and evaluating a model by providing pre-trained models. To use computer vision, you must create an Azure resource. The use of computer vision involves inferencing.

Which three parts of the machine learning process does the Azure AI Vision eliminate the need for? Each correct answer presents part of the solution. Azure resource provisioning choosing a model evaluating a model inferencing training a model

a webpage an existing FAQ document manually entered data NOTES: A webpage or an existing document, such as a text file containing question and answer pairs, can be used to generate a knowledge base. You can also manually enter the knowledge base question-and-answer pairs. You cannot directly use an image or an audio file to import a knowledge base.

Which three sources can be used to generate questions and answers for a knowledge base? Each correct answer presents a complete solution. Select all answers that apply. a webpage an audio file an existing FAQ document an image file manually entered data

Classification regression time-series forecasting NOTES: Time-series forecasting, regression, and classification are supervised machine learning models. Automated ML learning can predict categories or classes by using a classification algorithm, as well as numeric values as part of the regression algorithm, and at a future point in time by using time-series data. Inference pipeline is not a machine learning model. Clustering is unsupervised machine learning and automated ML only works with supervised learning algorithms

Which three supervised machine learning models can you train by using automated machine learning (automated ML) in the Azure Machine Learning studio? Each correct answer presents a complete solution. Select all answers that apply. Classification Clustering inference pipeline regression time-series forecasting

ISO 6391 Code, Language Name, and Score NOTES: Language Name, ISO 6391 Code, and Score are three values returned by the Language service of natural language processing (NLP) in Azure. Bounding box coordinates are returned by the Azure AI Vision services in Azure. Wikipedia URL is one of potential values returned by entity linking of entity recognition.

Which three values are returned by the language detection feature of the Azure AI Language service in Azure? Select all answers that apply. Bounding box coordinates ISO 6391 Code Language Name Score Wikipedia URL

the Speech service the Translator service NOTES: The Azure AI Speech service can be used to generate spoken audio from a text source for text-to-speech translation. The Azure AI Translator service directly supports text-to-text translation in more than 60 languages. Key phrase extraction, Conversational Language Understanding, and language detection are not used for language translation for text-to-text and speech-to-text translation.

Which two Azure AI Services features can be used to enable both text-to-text and speech-to-text between multiple languages? Each correct answer presents part of the solution. Select all answers that apply. Conversational Language Understanding language detection the Speech service the Translator service

optical character recognition (OCR) and spatial analysis NOTES: OCR and Spatial Analysis are part of the Azure AI Vision service. Sentiment analysis, entity recognition, and key phrase extraction are not part of the computer vision service.

Which two artificial intelligence (AI) workload features are part of the Azure AI Vision service? Each correct answer presents a complete solution. Select all answers that apply. entity recognition key phrase extraction optical character recognition (OCR) sentiment analysis spatial analysis

Create natural language and Understand natural language. NOTES: Azure OpenAI natural language models can take in natural language and generate responses. GPT models are excellent at both understanding and creating natural language.

Which two capabilities are examples of a GPT model? Each correct answer presents a complete solution. Select all answers that apply. Create natural language. Detect specific dialects of a language. Generate closed captions in real-time from a video. Synthesize speech. Understand natural language.

ID document model and invoice model NOTES: - The invoice model extracts key information from sales invoices and is suitable for extracting information from sales account documents. - The ID document model is optimized to analyze and extract key information from US driver's licenses and international passport biographical pages. - The business card model, receipt model, and language model are not suitable to extract information from passports or sales account documents.

Which two prebuilt models allow you to use the Azure AI Document Intelligence service to scan information from international passports and sales accounts? Each correct answer presents part of the solution. Select all answers that apply. business card model ID document model invoice model language model receipt model

accountability privacy and security NOTES: The accountability principle states that AI systems are designed to meet any ethical and legal standards that are applicable. The system must be designed to ensure that privacy of the healthcare data is of the highest importance, including anonymizing data where applicable. The fairness principle is applied to AI systems to ensure that users of the systems are treated fairly. The inclusiveness principle states that AI systems must empower people in a positive and engaging way.

Which two principles of responsible artificial intelligence (AI) are most important when designing an AI system to manage healthcare data? Each correct answer presents part of the solution. Select all answers that apply. accountability fairness inclusiveness privacy and security

celebrities and landmarks

Which two specialized domain models are supported by Azure AI Vision when categorizing an image? Each correct answer presents a complete solution. Select all answers that apply. celebrities image types landmarks people_ people_group

regression NOTES: The regression algorithms are used to predict numeric values. Clustering algorithms groups data points that have similar characteristics. Classification algorithms are used to predict the category to which an input value belongs. Unsupervised learning is a category of learning algorithms that includes clustering, but not regression or classification.

Which type machine learning algorithm predicts a numeric label associated with an item based on that item's features? Select only one answer. classification clustering regression unsupervised

semantic segmentation NOTES: Semantic segmentation provides the ability to classify individual pixels in an image depending on the object that they represent. The other answer choices also process images, but their outcomes are different.

Which type of artificial intelligence (AI) workload provides the ability to classify individual pixels in an image depending on the object that they represent? Select only one answer. image analysis image classification object detection semantic segmentation

clustering NOTES: A clustering algorithm is an example of unsupervised learning, which groups data points that have similar characteristics without relying on training and validating label predictions. Supervised learning is a category of learning algorithms that includes regression and classification, but not clustering. Classification and regression algorithms are examples of supervised machine learning.

Which type of machine learning algorithm finds the optimal way to split a dataset into groups without relying on training and validating label predictions? Select only one answer. classification clustering regression supervised

Azure AI Bot Service NOTES: Azure AI Bot Service provide a platform for conversational artificial intelligence (AI), which designates the ability of software agents to participate in a conversation. Azure AI Translator is part of Natural language processing (NLP), but it does not serve as a platform for conversational AI. Azure AI Vision deals with image processing. Azure AI Document Intelligence extracts information from scanned forms and invoices.

Which type of service provides a platform for conversational artificial intelligence (AI)? Select only one answer. Azure AI Bot Service Azure AI Document Intelligence Azure AI Vision Azure AI Translator

text-to-text NOTES: The Azure AI Translator service supports text-to-text translation, but it does not support speech-to-text, text-to-speech, or speech-to-speech translation.

Which type of translation does the Azure AI Translator service support? Select only one answer. speech-to-speech speech-to-text text-to-speech text-to-text

Accountability

You are building your AI solution within the framework of governance and organizational requirements that reflect defined legal and ethical standards. What responsible AI principle are you following? Fairness Accountability Inclusiveness Transparency Reliability and safety Privacy and security

positive

You are using Text Analytics Sentiment API to analyze the following sentence: "Peter was surprised and very happy to meet Sara in Paris." What sentiment value should you expect in the API response? positive negative surprise happy neutral

Feature selection Overall explanation During pre-processing, you need to work with data to select features that influence the label prediction. In this problem, features are the engine characteristics (power or volume), seat comforts, etc. They could help the ML model to predict the success of the new car model. In short, Feature selection helps us to narrow down the features that are important for our label prediction and discard all features that don't play or play a minimal role in a label prediction. As a result, our trained model and prediction will be more efficient. All other options are incorrect because they are parts of the different data processing events that are irrelevant to the pre-processing (Training set selection or Data for model evaluation selection) or too generic (Data selection or Data Classification).

You are working for a car dealership. Your boss asks you to provide him forecast information: will the new car model be successful or not. The new model has a variety of engine improvements, more comfortable seats, and a sunroof. You compiled the list of data about previous successful models with their characteristics and sales numbers. What should you do in the pre-processing data stage that would help you to predict the success of the new model? Data classification Training set selection Feature selection Data for model evaluation selection Data selection

No Overall explanation Your task is to provide a numeric prediction. You can achieve this by creating a regression model based on the historical sales data of the blue cars from previous quarters. Regression and Classification modeling types are two parts of Supervised machine learning. Only Clustering belongs to Unsupervised machine learning. If you choose the Unsupervised machine learning approach, you will not achieve your goal.

You are working for a car dealership. Your boss asks you to provide him information about how many blue cars he needs to order for the next quarter. You decide to create an ML model and choose an unsupervised machine learning approach. Will this help you to achieve your goal?

Class name Bounding box Probability score

You created a Custom Vision model. You want your model to detect trained objects on the photos. What information will you get about each object if you are using an object detection model? Image type Class name Bounding box Probability score Content name Image category

Inclusiveness Privacy & security Reliability & Safety Overall explanation Microsoft recognizes six principles of responsible AI: Fairness, Reliability and safety, Privacy and security, Transparency, Inclusiveness and Accountability. Therefore all other options are incorrect.

You created a Personal Virtual Assistant. Select all responsible AI principles that your solution must follow. Responsiveness Inclusiveness Privacy & security Reliability & Safety answerability dependability

4x4 Overall explanation The confusion matrix provides a tabulated view of predicted and actual values for each class. If we are predicting the classification for four classes, our confusion matrix will have a 4x4 size. All other options are incorrect.

You created a classification model with four possible classes. What will be the size of the confusion matrix? 3x3 4x4 2x2 6x6 10x10

No Overall explanation You need to normalize your numeric features. The process of normalization brings numeric features to a common scale. Feature engineering is the method of creating new features based on the existing ones.

You need to train and test your ML model. You prepare data for model training. Several of your numeric features have different scales. The first feature has a minimum value of 0.253 and a max of 0.987, the second one - from 12 to 124, and the last one - from 13545 to 56798. You need to bring them to a common scale. You decide to use feature engineering to address this problem. Does this decision help you to achieve your goal?

No Overall explanation You have to split your data into two sets: the first is for model training and the second for model testing. If you are using Automated machine learning, it automatically does that for you as part of data preparation and model training.

You need to train and test your model. You prepared data for model training. You decided to use all the data for model training and then for the model validation. Does this decision help you to achieve your goal?

a job NOTES: A job must be created in Machine Learning studio to use Machine Learning to train a regression model. A workspace must be created before you can access Machine Learning studio. An Azure container instance and an AKS cluster can be created as a deployment target, after training of a model is complete.

You need to use Azure Machine Learning to train a regression model. What should you create in Machine Learning studio? Select only one answer. a job a workspace an Azure container instance an Azure Kubernetes Service (AKS) cluster

Create a pipeline NOTES: Before you can start training a machine learning model, you must first create a pipeline in the Machine Learning designer. This is followed by adding a dataset, adding training modules, and eventually deploying a service.

You need to use the Azure Machine Learning designer to train a machine learning model. What should you do first in the Machine Learning designer? Select only one answer. Add a dataset. Add training modules. Create a pipeline. Deploy a service.

image description NOTES: Image description is not a capability included in the DALL-E model, therefore, it is not a use case that can be implemented by using DALL-E, while the other three capabilities are offered by DALL-E in Azure OpenAI.

You plan to develop an image processing solution that will use DALL-E as a generative AI model. Which capability is NOT supported by the DALL-E model? image description image editing image generation image variations

Deploy the model to an endpoint NOTES: You can deploy the best performing model for client applications to use over the internet by using an endpoint. Compute clusters are used to train the model and are created directly after you create a Machine Learning workspace. Before you can test the model's endpoint, you must deploy it first to an endpoint. Automated ML performs the validation automatically, so you do not need to split the dataset.

You train a regression model by using automated machine learning (automated ML) in the Azure Machine Learning studio. You review the best model summary. You need to publish the model for others to use from the internet. What should you do next? Select only one answer. Create a compute cluster. Deploy the model to an endpoint. Split the data into training and validation datasets. Test the deployed service.

Azure Bot Service

You want to build a personal virtual assistant. What service will you use to connect your assistant with various input channels and devices? Text Analytics LUIS Speech to Text Computer Vision Azure Bot Service QnA Maker

Classification model Overall explanation We are training the Classification model. In our case, we are using the historical data and predicting the price range category that a new phone belongs to. The "Price range" column is our target or label, and it has four classes: 0 (low cost), 1(medium cost), 2 (high cost), and 3 (very high cost). The model output value will be one of these four classes. The Regression model is wrong, the Regression model uses historical data for model training. But it predicts the output numeric value, not the class or classes. The Clustering model is wrong, the Clustering model clusters unlabeled data into groups based on some common properties. An unsupervised model uses unlabeled data.

Your company created a new mobile phone. You need to define a price range (0 - low cost to 3 - very high cost) for the phone. You collected technical and sales data for the phones on the market. Now you are ready to train your model. Here is your train dataset: Classification model Regression model Clustering model Unsupervised model

Color Data pre-processing involves various techniques, like feature selection, normalization or feature engineering, etc. During feature selection, we identify features that would help us with label prediction. And we discard the rest. In our dataset, the Color feature wouldn't correlate with the label due to the constant value of "black." We can safely remove this feature from the final dataset. All other options should be included in the training dataset.

Your company created a new mobile phone. You need to define a price range (0 - low cost to 3 - very high cost) for the phone. You collected technical and sales data for the phones on the market. Now you are ready to train your model. Here is your train dataset: What column will you discard from the final dataset during feature selection?

Price Range

Your company created a new mobile phone. You need to define a price range (0 - low cost to 3 - very high cost) for the phone. You collected technical and sales data for the phones on the market. Now you are ready to train your model. Here is your train dataset: What will be the label for this model? Battery power Price Range Clock speed Internal memory Color Dual sim

safety system NOTES: The safety system layer includes platform-level configurations and capabilities that help mitigate harm. For example, the Azure OpenAI service includes support for content filters that apply criteria to suppress prompts and responses based on the classification of content into four severity levels (safe, low, medium, and high) for four categories of potential harm (hate, sexual, violence, and self-harm).

At which layer can you apply content filters to suppress prompts and responses for a responsible generative AI solution? Select only one answer. metaprompt and grounding model safety system user experience

conversational and dictation NOTES: The Universal Language Model used by the speech-to-text API is optimized for conversational and dictation scenarios. The acoustic, language, and pronunciation scenarios require developing your own model.

For which two scenarios is the Universal Language Model used by the speech-to-text API optimized? Each correct answer presents a complete solution. Select all answers that apply. acoustic conversational dictation language pronunciation

Features are used to generate predictions for the label, which is compared to the actual label values. NOTES: In a regression machine learning algorithm, features are used to generate predictions for the label, which is compared to the actual label value. There is no direct comparison of features or labels between the validation and training datasets.

In a regression machine learning algorithm, how are features and labels handled in a validation dataset? Select only one answer. Features are compared to the feature values in a training dataset. Features are used to generate predictions for the label, which is compared to the actual label values. Labels are compared to the label values in a training dataset. The label is used to generate predictions for features, which are compared to the actual feature values.

Generative AI NOTES: Generative AI models offer the capability of generating images based on a prompt by using DALL-E models, such as generating images from natural language. The other AI capabilities are used in different contexts to achieve other goals.

Select the answer that correctly completes the sentence. [Answer choice] can return responses, such as natural language, images, or code, based on natural language input. Select only one answer. Computer vision Deep learning Generative AI Machine learning Reinforcement learning

QnA Maker Overall explanation You need to use the QnA Maker service. First, you need to provision the QnA Maker resource in your Azure subscription. After that, you can populate the newly created knowledge base using a frequently asked questions (FAQ) document. Azure Bot Service facilitates access to the knowledge base, but this service doesn't create a knowledge base. Custom vision service helps create your computer vision model, but this service doesn't create a knowledge base.  Text Analytics helps analyze text documents, detect documents' language, extract key phrases, determine entities, and provide sentiment analysis. This service doesn't create a knowledge base. Language Understanding Intelligent Service (LUIS) helps understand voice or text commands. This service doesn't create a knowledge base.

The customer service of your company spends a lot of time answering the same questions. They asked you to help them to automate this process. They provided you with a Microsoft Excel (*.xlsx) document with frequently asked questions and typical answers. What service will you use to create a knowledge base from this document? LUIS Text Analytics Custom vision QnA Maker Azure Bot Service

Kubernetes clusters Attached compute Compute clusters Compute Instances

What are the four types of Compute resources you can use in Azure Machine Learning Studio? Kubernetes clusters Classification clusters Attached compute AKS Cluster instances Compute clusters Compute Instances Automated ML instances

Impute missing values Normalize numeric features Feature selection Finding and removing data outliers Overall explanation After we ingest the data, we need to do a data preparation or transformation before supplying it for model training. There are four typical steps for data transformation such as Feature selection, Finding and removing data outliers, Impute missing values, and Normalize numeric features. Split data is coming after data transformation. ML algorithm selection data is coming after data transformation and Split Data steps.

What are the four typical steps of data transformation for model training? Split data Impute missing values ML algorithm selection Normalize numeric features Feature selection Finding and removing data outliers

Automated ML Notebooks Designer

What are the three main authoring tools on the Azure ML Studio home screen? Datasets Compute Automated ML Pipelines Notebooks Experiments Designer

Recall Precision Average Precision (AP) Overall explanation There are three main performance metrics for the Custom vision models: Precision, Recall, and Average Precision (AP). Precision defines the percentage of the class predictions that the model makes correct. For example, if the model predicts that ten images are bananas, and there are actually only seven bananas, the model precision is 70%. Recall defines the percentage of the class identification that the model makes correct. For example, if there are ten apple images, and the model identifies only eight, the model recall is 80%. Average Precision (AP) is the combined metrics of both Precision and Recall. Accuracy is a Classification model metric Number of Points is a Clustering model metric Mean Absolute Error (MAE) is a Regression model metric

What are the three metrics that help to evaluate Custom vision model performance? Recall Precision Accuracy Average Precision (AP) Mean Absolute Error (MAE) Number of Points

Coefficient of Determination Mean Absolute Error (MAE) Root Mean Squared Error (RMSE) Overall explanation Azure ML uses model evaluation for the measurement of the trained model accuracy. For regression models Evaluate Model module provides the following five metrics: Mean absolute error (MAE), Root mean squared error (RMSE), Relative absolute error (RAE), Relative squared error (RSE), and Coefficient of determination (R2). (RMSE) -regression model evaluation metrics. represents the square Root from the squared mean of the errors between predicted and actual values. (MAE) -regression model evaluation metrics. produces the score that measures how close the model is to the actual values — the lower score, the better the model performance. Coefficient of determination or R2 - regression model evaluation metrics. reflects the model performance: the closer R2 to 1 - the better the model fits the data.

What metrics does Azure ML use for the Evaluation of the regression models? Select all that apply. Coefficient of Determination Mean Absolute Error (MAE) Combined Evaluation Accuracy Number of Points Root Mean Squared Error (RMSE)

Categorize image Read the text in the image Identifies Landmarks Detects Objects Overall explanation Computer Vision service is one of the main areas of Artificial Intelligence. It belongs to the group of Azure Computer vision solutions such as Computer Vision service, Custom Vision Service, Face service, and Form Recognizer. Computer Vision service works with images. This service brings sense to the image pixels by using them as features for ML models. These predefined models help categorize and classify images, detect and recognize objects, tag, and identify them. Computer Vision can "read" a text in images in 25 languages and recognize landmarks. Find Similar Faces because the Find Similar Faces is the technique that is part of Face service and is not a Computer Vision service.

What tasks does Computer Vision Cognitive service include? Select all that apply. Processes forms Categorize image Read the text in the image Translator text Identifies Landmarks Detects Objects Find Similar Faces

face identification and face verification NOTES: - Face identification in the Azure AI Face service can address one-to-many matching of one face in an image to a set of faces in a secure repository. - Face verification has the capability for one-to-one matching of a face in an image to a single face from a secure repository or a photo to verify whether they are the same individual. - Face attributes, the find similar faces operation, and Azure AI Custom Vision do not verify the identity of a face.

When using the Azure AI Face service, what should you use to perform one-to-many or one-to-one face matching? Each correct answer presents a complete solution. Select all answers that apply. Custom Vision face attributes face identification face verification find similar faces

Azure OpenAI NOTES: Azure OpenAI is the only service capable of generating text that can be used in chat applications to create conversational experiences. The other workloads are Azure Cognitive Services used for different purposes, but not for generating text used in chat applications.

Which AI service can be integrated into chat applications and generate content in the form of text? Select only one answer. Azure AI Language Azure AI Metrics Advisor Azure AI Vision Azure OpenAI

named entity recognition NOTES: - Named entity recognition can identify and categorize entities in unstructured text, such as people, places, organizations, and quantities, and is suitable to support the development of an article recommendation system. - Key phrase extraction, Content Moderator, and the PII feature are not suited to entity recognition tasks to build a recommender system.

Which Azure AI Service for Language feature allows you to analyze written articles to extract information and concepts, such as people and locations, for classification purposes? Select only one answer. Azure AI Content Moderator key phrase extraction named entity recognition Personally Identifiable Information (PII) detection

sentiment analysis NOTES: Sentiment analysis provides sentiment labels, such as negative, neutral, and positive, based on a confidence score from text analysis. This makes it suitable for understanding user sentiment for product reviews. The named entity recognition, key phrase extraction, and language detection features cannot perform sentiment analysis for product reviews.

Which Azure AI Service for Language feature can be used to analyze online user reviews to identify whether users view a product positively or negatively? Select only one answer. key phrase extraction language detection named entity recognition sentiment analysis

confidence score NOTES: Each phrase returned by an image description task of the Azure AI Vision includes the confidence score. An endpoint and a key must be provided to access the Azure AI Vision service. Bounding box coordinates are returned by services such as object detection, but not image description.

Which additional piece of information is included with each phrase returned by an image description task of the Azure AI Vision? Select only one answer. bounding box coordinates confidence score endpointkey

object detection NOTES: - Detecting objects identifies common objects and, for each, returns bounding box coordinates. - Image categorization assigns a category to an image, but it does not return bounding box coordinates. - Tagging involves associating an image with metadata that summarizes the attributes of the image, but it does not return bounding box coordinates. - OCR detects printed and handwritten text in images, but it does not return bounding box coordinates.

Which analytical task of the Azure AI Vision service returns bounding box coordinates? image categorization object detection optical character recognition (OCR) tagging

Recall Overall explanation The confusion matrix (or error matrix) provides a tabulated view of predicted and actual values for each class. It is usually used as a performance assessment for classification models. A binary confusion matrix is divided into four squares that represent the following True positive (TP) True negative (TN) False positive (FP) False negative (FN) Recall metric defines how many positive cases that the model predicted are actually predicted right. We can calculate this metric using the following formula: TP/(TP+FN). Accuracy is incorrect: a formula for Accuracy metric calculation is (TP+TN)/Total number of cases. Precision is incorrect: a formula for Precision metric calculation is TP/(TP+FP). F1 Score is incorrect: a formula for F1 metric calculation is 2TP/(2TP+FP+FN). Selectivity is incorrect: this expression is for Selectivity (or true negative rate) metric calculation: TN/(TN+FP).

You created a classification model. Below is the confusion matrix for this model: What is the name of the metric that uses TP/(TP+FN) expression for its value calculation? Precision Selectivity Recall Accuracy F1 Score

Fairness

You created an AI solution that qualifies customers for a bank loan. The solution provides different results for the people living in cities and rural areas. What responsible AI principle does your solution violate? Fairness Accountability Inclusiveness Transparency Reliability and safety Privacy and security

Transparency Overall explanation Microsoft recognizes six principles of responsible AI: Fairness, Reliability and safety, Privacy and security, Transparency, Inclusiveness and Accountability. The principle of Transparency helps people to understand how to use AI solutions, their behavior, possibilities, and limitations.

You created an AI solution. Along with solution deployment, you provided information about the solution's possibilities and limitations. By providing this information, with what principle for responsible AI did you comply? Accountability Reliability and safety Inclusiveness Privacy and security Fairness Transparency

object detection

You have a set of images. Each image shows multiple vehicles. What allows you to identity different vehicle types in the same traffic monitoring image? Select only one answer. image classification linear regression object detection optical character recognition (OCR)

Semantic segmentation Overall explanation When the application processes images, it uses Semantic segmentation to classify pixels that belong to the particular object (in our case, flooded areas) and highlights them. Object detection is incorrect because Object detection helps to identify objects and their boundaries within the image. Image classification is incorrect because Image classification helps to classify images based on their content. Face detection is incorrect because Face detection is a Computer vision technique that helps detect and recognize people's faces. Image Analysis is incorrect because Image Analysis helps extract information from the images, tag them, and create a descriptive image summary.

You implement an aerial image processing application to identify the flooded areas. What common Computer Vision task is this application using? Object detection Semantic segmentation Face detection Image Analysis Image classification

Image classification Overall explanation You use your camera to capture a picture of the product. An application identifies this product utilizing the Image classification model and submits it for a search. The Image classification model helps to classify images based on their content. Object detection is incorrect because the Object detection model helps to identify objects and their boundaries within the image. Semantic segmentation is incorrect because the Semantic segmentation model helps classify pixels to the objects they belong to. Face detection is incorrect because Face detection is a Computer vision technique that helps detect and recognize people's faces. Image Analysis is incorrect because Image Analysis helps extract information from the images, tag them, and create a descriptive image summary.

You install a Visual product search application on your mobile. The application searches products based on the images that you capture by mobile camera. What Computer Vision common task this application uses for the product search? Object detection Semantic segmentation Face detection Image classification Image Analysis

Utterances Intents Entities Overall explanation For language model training, we need to provide the following key elements: Entities, Intents, and Utterance. We can achieve this by using the Azure Cognitive service LUIS portal. The Entity is the word or phrase that is the focus of the utterance, as the word "light" in the utterance "Turn the lights on." The Intent is the action or task that the user wants to execute. It reflects in utterance as a goal or purpose. We can define intent as "TurnOn" in the utterance "Turn the lights on." The Utterance is the user's input that your model needs to interpret, like "Turn the lights on" or "Turn on the lights".

You need to create a language model. What are the essential elements that you need to supply as data for your language model training? Select all that apply. Utterances Subjects Intents Entities Knowledge domains Verbs

Select Columns in Dataset Overall explanation After we bring data for model training or ingest data, the next stage is the Data transformation. Data transformation or data pre-processing usually includes the following steps: feature selection, data cleaning, and data normalization. In Azure ML Designer, we need to drag-and-drop the "Select Columns in Dataset" module from the Data Transformation section. Then on the right-side panel, we can select all the features we want to use for the model training.

You need to create a new pipeline to train a regression model using Azure ML Designer. You ingest your data for the model and drop it on the canvas. What module would you typically drag-and-drop next on the canvas? Train Model Clean Missing Data Normalize Data Split Data Select Columns in Dataset

logistic regression NOTES: Multiple linear regression models a relationship between two or more features and a single label. Linear regression uses a single feature. Logistic regression is a type of classification model, which returns either a Boolean value or a categorical decision. Hierarchical clustering groups data points that have similar characteristics.

You need to identify numerical values that represent the probability of humans developing diabetes based on age and body fat percentage. Which type of machine learning model should you use? Select only one answer. hierarchical clustering linear regression logistic regression multiple linear regression