Microsoft Azure AI Fundamentals AI-900 Exam Guide E-Book

Aaron Guilmette

• Herausgeber: Packt Publishing
• Kategorie: Fachliteratur
• Sprache: Englisch

Seitenzahl: 328

Veröffentlichungsjahr: 2024

Beschreibung

The AI-900 exam helps you take your first step into an AI-shaped future. Regardless of your technical background, this book will help you test your understanding of the key AI-related topics and tools used to develop AI solutions in Azure cloud.
This exam guide focuses on AI workloads, including natural language processing (NLP) and large language models (LLMs). You’ll explore Microsoft’s responsible AI principles like safety and accountability. Then, you’ll cover the basics of machine learning (ML), including classification and deep learning, and learn how to use training and validation datasets with Azure ML. Using Azure AI Vision, face detection, and Video Indexer services, you’ll get up to speed with computer vision-related topics like image classification, object detection, and facial detection. Later chapters cover NLP features such as key phrase extraction, sentiment analysis, and speech processing using Azure AI Language, speech, and translator services. The book also guides you through identifying GenAI models and leveraging Azure OpenAI Service for content generation. At the end of each chapter, you’ll find chapter review questions with answers, provided as an online resource.
By the end of this exam guide, you’ll be able to work with AI solutions in Azure and pass the AI-900 exam using the online exam prep resources.

Leseprobe

Microsoft Azure AI Fundamentals AI-900 Exam Guide

Gain proficiency in Azure AI and machine learning concepts and services to excel in the AI-900 exam

Aaron Guilmette

Steve Miles

Microsoft Azure AI Fundamentals AI-900 Exam Guide

Copyright © 2024 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the authors, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.

Associate Group Product Manager: Niranjan Naikwadi

Publishing Product Manager: Surbhi Suman

Book Project Manager: Ashwin Kharwa

Senior Editor: Divya Vijayan

Technical Editor: Irfa Ansari

Copy Editor: Safis Editing

Proofreader: Divya Vijayan

Indexer: Tejal Daruwale Soni

Production Designer: Prafulla Nikalje

Senior DevRel Marketing Executive: Vinishka Kalra, Kunal Raj

First published: May 2024

Production reference: 1170524

Published by Packt Publishing Ltd.

Grosvenor House
11 St Paul’s Square
Birmingham
B3 1RB, UK

ISBN 978-1-83588-566-6

www.packtpub.com

I’d like to thank my long-suffering girlfriend, Christine, who has put up with my book deadlines for the last 14 books. She’s the real hero. Also, I’d like to thank my kids—Liberty, Hudson, Anderson, Glory, and Victory—without them, I’d probably be able to retire sooner to a tropical location with umbrella drinks.

– Aaron Guilmette


I’d like to thank the people who have supported me professionally and at home, especially my wife, Pippa, aka Mrs Smiles, and my family. Also, apologies to my two four-legged family members, Henry and Lilly, for their missed walks while I have been locked away at evenings and weekends writing.

– Steve Miles

Foreword

Artificial intelligence (AI) is transforming the world in unprecedented ways. From enhancing customer experiences to automating business processes, AI enables organizations to achieve more with fewer resources and less time. However, AI is not a magic bullet that can solve any problem without human guidance and expertise. To harness the full potential of AI, you need to understand its core concepts, principles, and applications, as well as the ethical and social implications of using it.

This is where Microsoft Azure AI comes in. Azure AI offers a comprehensive set of cloud services and tools that enable you to build, deploy, and manage AI solutions at scale. Whether you want to create intelligent chatbots, analyze images and videos, generate natural language, or make predictions based on data, Azure AI provides the necessary services. It also integrates with other Azure services, such as Azure Data, Azure DevOps, and Azure Security, to offer a seamless and secure end-to-end AI development life cycle.

But how do you kickstart your journey with Azure AI? How do you choose the appropriate service for your specific use case? How do you design, implement, and optimize your AI solutions? How do you ensure that your AI solutions are ethical, responsible, and trustworthy? These are some of the questions that this book, Microsoft Azure AI Fundamentals (AI-900) Exam Guide, aims to address.

I’ve collaborated with Steve Miles for several years, and have worked as a technical reviewer for all his books, and enjoying each one. Steve’s books are packed with amazing content, going above and beyond in clarifying complex matters, having detailed diagrams for clarity, and bringing in a lot of his own practical knowledge and great skill set. In my job as a Microsoft Technical Trainer, I often refer to his material when learners ask me for additional content for certification preparation.

This book serves as a comprehensive guide to prepare you for the Microsoft Azure AI Fundamentals certification exam (AI-900). This exam is designed to assess your foundational knowledge of AI and machine learning concepts, as well as your ability to use Azure AI services to implement AI solutions. By passing this exam, you will demonstrate your competence and confidence in using Azure AI, earning you a valuable credential that can boost your career prospects and credibility.

This book covers all the topics and objectives of the AI-900 exam, with clear explanations, practical examples, and self-assessment questions. You will learn about the following topics:

The principles and concepts of AI and machine learning, such as supervised and unsupervised learning, deep learning, computer vision, natural language processing, conversational AI, and anomaly detectionThe features and capabilities of Azure AI services, such as Azure Cognitive Services, Azure AI Bot Service, Azure Machine Learning, and Azure Cognitive SearchThe best practices and considerations for designing, implementing, and optimizing AI solutions using Azure AI services, such as choosing the right service, data preparation, model training and deployment, performance monitoring, and security and privacyThe ethical and social aspects of AI, such as the principles of responsible AI, the risks and challenges of AI, and the tools and frameworks for ensuring fairness, reliability, accountability, and transparency of AI solutions

By reading this book, you will not only prepare for the AI-900 exam but also establish a solid foundation of Azure AI for your future AI projects and endeavors. You will also cultivate a critical and responsible mindset that enables you to leverage AI for positive outcomes and avoid its pitfalls.

AI is a powerful and exciting field that offers endless possibilities and opportunities. With Azure AI, you can unleash your creativity and innovation to develop AI solutions that can make a positive impact on your organization and society. Whether you are a beginner or an experienced professional, this book will help you in achieving your Azure AI goals and aspirations.

I hope you enjoy reading this book and learning from it as much as I did. I wish you all the best in your AI-900 exam and for your Azure AI journey.

Peter De Tender

Microsoft Technical Trainer,

Microsoft Corp, Redmond.

Contributors

About the authors

Aaron Guilmette is a principal architect at Planet Technologies, an award-winning Microsoft Partner focused on dragging public sector public sector customers into the modern era. Previously, he worked at Microsoft as a senior program manager for Microsoft 365 Customer Experience. As the author of over a dozen IT books, he specializes in identity, messaging, and automation technologies. When he’s not writing books or tools for his customers, trying to teach one of his kids to drive, or making tacos with his girlfriend, Aaron can be found tinkering with cars. You can visit his blog at https://aka.ms/aaronblog or connect with him on LinkedIn at https://www.linkedin.com/in/aaronguilmette.

Steve Miles is CTO at Westcoast Cloud, part of a multi-billion turnover IT distributor based in the UK and Ireland. Steve is a Microsoft Most Valuable Professional (MVP), Microsoft Certified Trainer (MCT), and an Alibaba Cloud MVP. He has 25+ years of technology experience and a previous military career in engineering, signals, and communications. Among other books, Steve is the author of the #1 Amazon best-selling AZ-900 certification title Microsoft Azure Fundamentals and Beyond. His books can be found on his author profile on Amazon at https://www.amazon.com/stores/Steve-Miles/author/B09NDJ1RC8.

Like Aaron, Steve is also a petrolhead, and can also be found tinkering with cars when he is not writing. You can connect with him on LinkedIn at https://www.linkedin.com/in/stevemiles70/.

About the reviewers

Peter De Tender has an extensive background in architecting, deploying, managing, and training Microsoft technologies, dating back to Windows NT4 Server in 1996, all the way to the latest and modern cloud solutions available in Azure today. With a passion for cloud architecture, DevOps, app modernization, and AI solutions, Peter always has a story to share on how to optimize your enterprise-ready cloud workloads.

Peter was an Azure MVP for 5 years, has been an MCT for 13+ years, and is still actively involved in the community as a public speaker, technical writer, book author, and publisher.

You can follow Peter on Twitter/X (@pdtit) and read his technical blog adventures at http://www.007ffflearning.com.

Jetro WILS is a cloud and information security advisor who began providing managed IT services as a teenager. He’s a certified Microsoft cybersecurity architect, Azure solutions architect, and MCT.

“For 18 years, I’ve been active in various tech companies in Belgium. From developer to business analyst to product manager to cloud specialist, I’ve experienced digital evolution first-hand. I’ve seen the rise of cloud technology fundamentally change business operations, yet many organizations struggle to adopt the cloud securely. Also, Europe adds more legislation yearly, making it harder to maintain compliance.”

Jetro is the founder of BlueDragon Security (www.bluedragonsecurity.com), where he helps organizations operate safely in the cloud.

Syed Mohamed Thameem Nizamudeen is a pioneer and subject matter expert (SME) in application modernization and cloud computing. He holds multiple industry certifications in Azure, AWS, GCP, Oracle, PMP, and CSM. Syed has over a decade of experience in product development, architecture, scalability, and modernization. He has worked with Oracle, leading top technology advisory firms like PwC and Ernst & Young, and other contemporary application enterprises. Syed is a skilled and dedicated technology management executive, leveraging his 15+ years of experience in designing and developing high-demand On-Premise/Commercial Off-The-Shelf Software/SaaS solutions.

I am honored to have served as a technical reviewer for the AI Exam Guide. This role has allowed me to contribute to shaping content that is both enlightening and essential for anyone looking to navigate the complex world of artificial intelligence. I extend my heartfelt thanks to the authors and editorial team for their collaborative spirit and dedication to excellence.

Table of Contents

Preface

Part 1: Identify Features of Common AI Workloads

1

Identify Features of Common AI Workloads

Making the Most Out of this Book – Your Certification and Beyond

Identify features of data monitoring and anomaly detection workloads

Identify features of content moderation and personalization workloads

Identify computer vision workloads

Identify natural language processing workloads

Identify document intelligence workloads

Summary

Exam Readiness Drill

Working On Timing

2

Identify the Guiding Principles for Responsible AI

Understanding ethical principles

Describe considerations for accountability

Describe considerations for inclusiveness

Describe considerations for reliability and safety

Understand explainable principles

Describe considerations for fairness

Describe considerations for transparency

Describe considerations for privacy and security

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

Part 2: Describe the Fundamental Principles of Machine Learning on Azure

3

Identify Common Machine Learning Techniques

Understanding machine learning terminology

Training

Inferencing

Identify regression machine learning scenarios

Example

Evaluation metrics

Applications

Identify classification machine learning scenarios

Binary classification

Multiclass classification

Identify clustering machine learning scenarios

Example

Evaluation metrics

Applications

Identify features of deep learning techniques

Example

Applications

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

4

Describe Core Machine Learning Concepts

Identify features and labels in a dataset for machine learning

Identifying features in a dataset

Identifying labels in a dataset

Describe how training and validation datasets are used in machine learning

Training set

Validation set

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

5

Describe Azure Machine Learning Capabilities

What is Azure ML?

Describe capabilities of AutoML

AutoML use cases

Training, validation, and test scenarios

Feature engineering

Ensemble models

Describe data and compute services for data science and machine learning

Compute

Data

Datastore

Environments

Model

Workspaces

Subscription

Storage account

Key Vault

Application Insights

Container Registry

Describe model management and deployment capabilities in Azure ML

Model management and deployment capabilities

MLOps

Build a machine learning model in Azure ML

Creating a machine learning workspace

Using AutoML to train a model

Reviewing and selecting the best model

Deploying and testing the model

Testing the deployed model service

Teardown

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

Part 3: Describe Features of Computer Vision Workloads on Azure

6

Identify Common Types of Computer Vision Solutions

Introduction to CV solutions

Image processing

CV ML

Identify features of image classification solutions

Identify features of object detection solutions

Identify features of OCR solutions

Identify features of facial detection and facial analysis solutions

Facial detection

Facial analysis

Facial recognition

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

7

Identify Azure Tools and Services for Computer Vision Tasks

Technical requirements

Describe capabilities of the Azure AI Vision service

Image classification

Object detection

OCR solutions

Describe the capabilities of the Azure AI Face service

Getting started

Facial detection

Responsible AI

Describe capabilities of the Azure AI Video Indexer service

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

Part 4: Describe Features of Natural Language Processing (NLP) Workloads on Azure

8

Identify Features of Common NLP Workload Scenarios

Introduction to NLP

NLP concepts

NLP scenarios

Identify features and uses for key phrase extraction

Identify features and uses for entity recognition

Identify features and uses for sentiment analysis

Identify features and uses for language modeling

Conversational language understanding (CLU)

Conversational AI

Identify features and uses for speech recognition and synthesis

Speech recognition

Speech synthesis

Identify features and uses for translation

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

9

Identify Azure Tools and Services for NLP Workloads

Technical requirements

Describe capabilities of the Azure AI Language service

Text analysis

Conversational language understanding

Question-answering

Azure AI Language Studio

Describe capabilities of the Azure AI Speech service

Azure AI Speech Studio

Describe capabilities of the Azure AI Translator service

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

Part 5: Describe Features of Generative AI Workloads on Azure

10

Identify Features of Generative AI Solutions

What is Generative AI?

Identify Features of Generative AI models

What’s a transformer model and how does it work?

How does generative AI put all this together?

Identify common scenarios for generative AI

Image generation

Text generation

Music creation

Synthetic data generation

Code generation

Voice generation and transformation

Drug discovery and chemical synthesis

Personalized content and recommendation systems

Maintenance analysis

Copilots

Deepfake creation and detection

Quality control

Identify Responsible AI considerations for generative AI

Identify

Measure

Mitigate

Operate

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

11

Identify Capabilities of Azure OpenAI Service

What is Azure OpenAI Service?

What’s included?

What’s the difference between Azure AI and Azure OpenAI services?

Accessing Azure OpenAI services

Describe natural language generation capabilities of Azure OpenAI Service

Describe code generation capabilities of Azure OpenAI Service

Describe image generation capabilities of Azure OpenAI Service

Summary

Exam Readiness Drill – Chapter Review Questions

Exam Readiness Drill

Working On Timing

12

Accessing the Online Practice Resources

How to Access These Resources

Purchased from Packt Store (packtpub.com)

Packt+ Subscription

Purchased from Amazon and Other Sources

Troubleshooting Tips

Practice Resources – A Quick Tour

A Clean, Simple Cert Practice Experience

Practice Questions

Flashcards

Exam Tips

Chapter Review Questions

Share Feedback

Back to the Book

Index

Other Books You May Enjoy

Preface

The AI-900 certification exam, also known as the Microsoft Azure AI Fundamentals exam, is designed to validate foundational knowledge of artificial intelligence (AI) concepts and how they are implemented in Microsoft Azure.

The AI-900 exam has been updated a few times to include new technologies as they emerge and enter the Azure space—and this edition of the exam is no different. AI-900 now includes a focus on the new OpenAI services available as part of an Azure subscription.

Who this book is for

This book is intended for individuals who are interested in gaining a basic understanding of AI and its applications in Azure but who may not have extensive technical experience in the field. This includes business stakeholders, decision-makers, and technical professionals who are new to AI technologies.

The content in this book assumes you have no knowledge of any machine learning or AI concepts (though it certainly helps with understanding some of the more complex topics).

What this book covers

Chapter 1, Identify Features of Common AI Workloads, introduces some of the basic concepts of AI in the Azure platform space.

Chapter 2, Identify the Guiding Principles for Responsible AI, explains Microsoft’s principles for responsible AI, such as transparency and inclusiveness.

Chapter 3, Identify Common Machine Learning Techniques, explores machine learning techniques, such as clustering and regression.

Chapter 4, Describe Core Machine Learning Concepts, expands on the concepts of machine learning techniques with explanations of features, labels, training, and validation.

Chapter 5, Describe Azure Machine Learning Capabilities, focuses on the power of automated machine learning (AutoML) as well as the functional resources necessary for enabling machine learning in Azure.

Chapter 6, Identify Common Types of Computer Vision Solutions, introduces the concepts behind computer vision, such as optical character recognition and object detection.

Chapter 7, Identify Azure Tools and Services for Computer Vision Tasks, expands on the basics of computer vision with services such as Azure AI Vision and Azure AI Face.

Chapter 8, Identify Features of Common NLP Workload Scenarios, introduces the core foundational workload uses for natural language processing, such as sentiment analysis, translation, and key phrase extraction.

Chapter 9, Identify Azure Tools and Services for NLP Workloads, provides information about Azure’s natural language processing solutions, such as the Azure AI Translator and Azure AI Language services.

Chapter 10, Identify Features of Generative AI Solutions, explains the broad features and use cases for generative AI models.

Chapter 11, Identify Capabilities of Azure OpenAI Service, highlights the powerful features of Azure OpenAI Service, including text content and image generation.

Chapter 12, Accessing the Online Practice Resources.

To get the most out of this book

To make the most of your studying experience, we recommend the following components:

Azure tenant with free trial subscriptions (https://azure.microsoft.com/en-us/free/ai-services/)Register for OpenAI access (https://aka.ms/oai/access)

If you are using the digital version of this book, we advise you to type the code yourself or access the code from the book’s GitHub repository (a link is available in the next section). Doing so will help you avoid any potential errors related to the copying and pasting of code.

Online Practice Resources

With this book, you will unlock unlimited access to our online exam-prep platform (Figure 0.1). This is your place to practice everything you learn in the book.

How to access the resources

To learn how to access these resources, head over to Chapter 12, Accessing the Online Resources, at the end of the book.

Figure 0.1: Dashboard interface of the online practice resources

Download the example code files

You can download the example code files for this book from GitHub at https://github.com/PacktPublishing/Microsoft-Azure-AI-Fundamentals-AI-900-Exam-Guide. If there’s an update to the code, it will be updated in the GitHub repository.

We also have other code bundles from our rich catalog of books and videos available at https://github.com/PacktPublishing/. Check them out!

Conventions used

There are a number of text conventions used throughout this book.

Code in text: Indicates code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles. Here is an example: “Mount the downloaded WebStorm-10*.dmg disk image file as another disk in your system.”

A block of code is set as follows:

[     {         "recognitionModel": "recognition_01",         "faceRectangle": {         "width": 144,         "height": 209,         "left": 305,         "top": 473 },

Bold: Indicates a new term, an important word, or words that you see onscreen. For instance, words in menus or dialog boxes appear in bold. Here is an example: “Select System info from the Administration panel.”

Tips or important notes

Appear like this.

Get in touch

Feedback from our readers is always welcome.

General feedback: If you have questions about any aspect of this book, email us at [email protected] and mention the book title in the subject of your message.

Errata: Although we have taken every care to ensure the accuracy of our content, mistakes do happen. If you have found a mistake in this book, we would be grateful if you would report this to us. Please visit www.packtpub.com/support/errata and fill in the form.

Piracy: If you come across any illegal copies of our works in any form on the internet, we would be grateful if you would provide us with the location address or website name. Please contact us at [email protected] with a link to the material.

If you are interested in becoming an author: If there is a topic that you have expertise in and you are interested in either writing or contributing to a book, please visit authors.packtpub.com.

Share Your Thoughts

Once you’ve read Microsoft Azure AI Fundamentals AI-900 Exam Guide, we’d love to hear your thoughts! Please click here to go straight to the Amazon review page for this book and share your feedback.

Your review is important to us and the tech community and will help us make sure we’re delivering excellent quality content.

Download a free PDF copy of this book

Thanks for purchasing this book!

Do you like to read on the go but are unable to carry your print books everywhere?

Is your eBook purchase not compatible with the device of your choice?

Don’t worry, now with every Packt book you get a DRM-free PDF version of that book at no cost.

Read anywhere, any place, on any device. Search, copy, and paste code from your favorite technical books directly into your application.

The perks don’t stop there, you can get exclusive access to discounts, newsletters, and great free content in your inbox daily

Follow these simple steps to get the benefits:

Scan the QR code or visit the link below

https://packt.link/free-ebook/9781835885666

Submit your proof of purchaseThat’s it! We’ll send your free PDF and other benefits to your email directly

Part 1: Identify Features of Common AI Workloads

In this first part of the book, you will be introduced to foundational concepts of artificial intelligence (AI) workloads as well as Microsoft’s principles for responsible AI development.

This part includes the following chapters:

Chapter 1, Identify Features of Common AI Workloads Chapter 2, Identify the Guiding Principles for Responsible AI

1

Identify Features of Common AI Workloads

Welcome to the world of artificial intelligence (AI)!

When you think of AI, what comes to mind? If you’ve watched sci-fi movies, your mind might conjure images of walking and talking human-like androids from movies such as Blade Runner or Terminator. Or perhaps you think of just a voice, such as the HAL 9000 in 2001: A Space Odyssey or Tony Stark’s seemingly all-knowing computational butler, Jarvis, as featured in Iron Man.

While that future is still a ways off, there are some pretty amazing things that AI can do right now. Artificial intelligence is software technology that imitates certain human capabilities, such as interpreting images and language or predicting outcomes of scenarios based on historical data and patterns.

In this book, you’ll learn about the broad range of AI technologies and capabilities that are available inside the Microsoft Azure platform. And, throughout this book, you’ll be exposed to examples, demos, and labs that show you how Azure AI services can be used to help address both simple and complex business scenarios.

The objectives and skills we’ll cover in this chapter include the following:

Identify features of data monitoring and anomaly detection workloadsIdentify features of content moderation and personalization workloadsIdentify computer vision workloadsIdentify natural language processing workloadsIdentify Knowledge Mining workloadsIdentify document intelligence workloadsIdentify features of generative AI workloads

By the end of this chapter, you should be able to discuss the features and capabilities of AI workloads and capabilities available in Microsoft Azure.

Get ready for your first steps on this exciting journey!

Note

Azure Cognitive Services was rolled up into the new Azure AI Services family branding. Neither pricing nor capabilities have changed as a result of the branding change.

Before we move on to the topics of this chapter, have a look at the following section.

Making the Most Out of this Book – Your Certification and Beyond

This book and its accompanying online resources are designed to be a complete preparation tool for your Microsoft Azure AI Fundamentals AI-900 Exam.

The book is written in a way that you can apply everything you’ve learned here even after your certification. The online practice resources that come with this book (Figure 1.1) are designed to improve your test-taking skills. They are loaded with timed mock exams, interactive flashcards, and exam tips to help you work on your exam readiness from now till your test day.

Before You Proceed

To learn how to access these resources, head over to Chapter 12, Accessing the Online Practice Resources, at the end of the book.

Figure 1.1 – Dashboard interface of the online practice resources

Read each section thoroughly.

Here are some tips on how to make the most out of this book so that you can clear your certification and retain your knowledge beyond your exam:

Make ample notes: You can use your favorite online note-taking tool or use a physical notebook. The free online resources also give you access to an online version of this book. Click the BACK TO THE BOOK link from the Dashboard to access the book in Packt Reader. You can highlight specific sections of the book there.Chapter Review Questions: At the end of this chapter, you’ll find a link to review questions for this chapter. These are designed to test your knowledge of the chapter. Aim to score at least 75% before moving on to the next chapter. You’ll find detailed instructions on how to make the most of these questions at the end of this chapter in the Exam Readiness Drill – Chapter Review Questions section. That way, you improve your exam-taking skills after each chapter, rather than at the end.Flashcards: After you’ve gone through the book and scored 75% more in each of the chapter review questions, start reviewing the online flashcards. They will help you memorize key concepts.Mock Exams: Solve the mock exams that come with the book till your exam day. If you get some answers wrong, go back to the book and revisit the concepts you’re weak in.Exam Tips: Review these from time to time to improve your exam readiness even further.

Identify features of data monitoring and anomaly detection workloads

Anomaly Detector is an AI service equipped with a suite of application programming interfaces (APIs) designed to empower users in monitoring and identifying anomalies within their time series data, even with limited machine learning (ML) expertise. Whether you require batch validation (a method for checking a model’s efficacy using a subset of training data) or real-time inference (making predictions using machine learning models), Anomaly Detector has you covered.

What’s time series data?

Time series data refers to a type of data where observations are collected or recorded over regular intervals of time. These observations are typically ordered chronologically, with each data point associated with a specific time stamp. Time series data is common in various areas such as finance, economics, weather forecasting, and sales.

This service offers two primary functionalities:

Univariate anomaly detectionMultivariate anomaly detection

Univariate anomaly detection allows users to identify anomalies in a single variable, such as revenue or cost, without the need for extensive ML knowledge. The model selection process is automated based on patterns in the data itself, ensuring optimal performance regardless of industry, scenario, or data volume. By leveraging time series data, the API establishes boundaries for anomaly detection, determines expected values, and then identifies anomalous data points.

On the other hand, multivariate anomaly detection APIs enable developers to integrate advanced AI capabilities for detecting anomalies across groups of metrics, eliminating the requirement for ML expertise or previously labeled data. These APIs automatically account for dependencies and inter-correlations between signals, crucial for safeguarding complex systems such as software applications, servers, factory machines, and spacecraft from failures.

In the real world, multivariate anomaly detection is frequently used to help identify things such as credit card transaction fraud. By training on data such as places you normally shop, locations you normally travel, and the average size of transactions, financial institutions can detect when your credit card has been compromised and alert you right away.

Exam tip

Multivariate anomaly detection can correlate up to 300 signals.

When deviations occur beyond the usual range of signal interactions, the multivariate anomaly detection feature acts as a seasoned expert, promptly identifying anomalies. The underlying AI models are trained and tailored using user data to address the unique requirements of their business. With the addition of these APIs, developers can seamlessly integrate multivariate time series anomaly detection capabilities into predictive maintenance solutions, AIOps monitoring solutions for complex enterprise software, or business intelligence tools.

Next, we’ll look at features of content moderation and personalization.

Identify features of content moderation and personalization workloads

Using AI to monitor and moderate content is also a growing task area. Content moderation refers to the process of screening user-generated content to ensure it adheres to certain standards or guidelines. The moderation APIs give AI developers ways to submit content for programmatic evaluation.

Note

The legacy Azure Content Moderator has been slated for retirement in February 2027. While it is still available, Microsoft recommends developers start switching to Azure AI Content Safety, which includes more robust features.

Azure AI Content Safety offers a comprehensive solution for detecting harmful content, encompassing both user-generated and AI-generated material across applications and services. This suite includes text and image APIs, along with an interactive Content Safety Studio, providing developers with the tools to identify and mitigate potentially harmful content effectively.

Content moderation plays a critical role in various industries, ensuring compliance with regulations and maintaining a safe environment for users. Scenarios where content moderation services are essential include online marketplaces, gaming companies, social messaging platforms, enterprise media, and K-12 education solutions.

The service offers different types of analysis through its APIs, including text and image analysis for sexual content, violence, hate speech, and self-harm, along with newer functionalities such as jailbreak risk detection and protected material text detection.

Azure AI Content Safety Studio serves as a powerful online tool for handling offensive or risky content, equipped with advanced content moderation ML models. It allows users to customize workflows, upload their own content, and utilize pre-built AI models and blocklists provided by Microsoft, ensuring comprehensive coverage of harmful content.

With Content Safety Studio, businesses can establish moderation workflows, continuously monitor and improve content moderation performance, and meet the specific content requirements of their industries. The platform simplifies operations, enabling quick validation of different solutions and facilitating efficient content moderation without the need for extensive model development.

Additionally, you can configure filters and thresholds for different types of potentially harmful content. Depending on the scenarios, you can tweak the filter settings to be more or less permissive, and then test samples of content against the filters to ensure the right type of content is getting blocked, as shown in Figure 1.1:

Figure 1.2 – Evaluating text in Content Safety Studio

Key features of Content Safety Studio include the ability to moderate text and image content, monitor online activity, and access detailed response information such as category distribution, latency, and error detection. Its user-friendly interface empowers developers to configure content filters, manage blocklists of prohibited terms, and implement moderation tools directly into their applications, streamlining workflow processes and enhancing content safety measures.

Further exploration

You can take a test drive of Content Safe Studio (part of Azure AI Cognitive Services) here:

https://contentsafety.cognitive.azure.com/.

Azure also contains personalization services called the AI Personalizer. Like content moderation, the personalizer analyzes content. However, instead of making decisions about a content’s safety or possible offensiveness, it’s used to make predictions on user behaviors, such as the following:

Using recent choice data or items already viewed, will the customer make a purchase?Based on the things viewed already, what other products or articles might be interesting to the user?Where should an advertisement be placed for optimal exposure?How should a popup notification be deployed to maximize visibility or response?Are there other data points available from partner or affiliate organizations that could be used to help make better decisions?

The Azure Personalizer works through the use of reinforcement learning (a type of machine learning) by assigning a point value (reward) for actions or choices that a user can make based on the current context of their session (location, items viewed, device, previous browsing history, or other information that the service has been able to gather). With this information, the personalizer service can then make automated decisions on what content to present to the user to encourage a particular choice or response.

Note

The Azure Personalizer has been deprecated and will be taken offline in October 2026. As of this writing, new Personalizer resources can no longer be created.

For more information on Azure Personalizer service, see the following: https://learn.microsoft.com/en-us/azure/ai-services/personalizer/what-is-personalizer.

In the next section, we’ll review the high-level features of computer vision.

Identify computer vision workloads

Azure’s AI Vision (also referred to as computer vision) service offers access to cutting-edge algorithms designed to process images and provide relevant information based on specific visual features.

The service boasts several key functionalities:

Optical character recognition (OCR): This feature enables the extraction of text from images, including both printed and handwritten text from various sources such as photos and documents. Utilizing deep-learning-based models, the OCR service works across different surfaces and backgrounds, including business documents, invoices, receipts, posters, business cards, letters, and whiteboards. Additionally, it supports multiple languages for extracting printed text.Image analysis: The image analysis service extracts a wide range of visual features from images, including objects, faces, adult content, and automatically generated text descriptions.Face recognition: The face service provides AI algorithms for detecting, recognizing, and analyzing human faces within images. This capability finds application in various scenarios such as identification, touchless access control, and privacy protection through face blurring.Spatial analysis: The spatial analysis service analyzes the presence and movement of individuals within video feeds, generating events that can trigger responses in other systems.

Some of the common uses or features for computer vision might include generating captions of images (such as “dog retrieves frisbee in a park”) or providing an analysis for objects detected in an image (such as a dog, a frisbee, the outdoors, a tree, or grass). When detecting objects, the Azure AI Vision service also returns information on confidence (or how sure the service is of the detection).

Azure AI Vision is particularly relevant for digital asset management (DAM) scenarios. DAM involves organizing, storing, and retrieving rich media assets while managing digital rights and permissions. For example, a stock photography service may design a digital asset management system to apply AI-generated descriptions of photographs, enabling customers to search for images.

Overall, Azure AI Vision provides a robust platform for leveraging advanced image processing capabilities to enhance digital asset management and various other applications.

Next, we’ll look at the features and capabilities for processing language in Azure.

Identify natural language processing workloads

Natural language processing (NLP) is a branch of AI focused on comprehending and responding to human language. The goal of NLP is to allow computers to interpret text similarly to humans and provide realistic dialogue and responses. Several popular and common consumer AI-based services, such as OpenAI’s ChatGPT or predictive text on a smartphone, use natural language processing to help understand input and context.

NLP is a building block for many other AI services, such as text analytics, which extract information from unstructured text. Examples of NLP applications include sentiment analysis for product marketing campaigns on social media, document summarization in a catalog search application, and extracting brands and company names from text.

Azure AI Language, a cloud-based service, offers tools for understanding and analyzing text. It features sentiment analysis, key phrase identification, text summarization, and conversational language understanding capabilities. These capabilities can help organizations process volumes of internal and customer-generated data to help make information more accessible and digestible as well as highlight patterns or anomalies that need to be addressed.

Azure AI Language has the following features and sub-services:

Text Analytics: This feature provides sentiment analysis, key phrase extraction, named entity recognition (identifying and categorizing items of interest), and language detection. It allows applications to understand the context, evaluate the sentiment of written text, identify important concepts, and recognize entities such as people, locations, and organizations.Translator: This feature offers real-time, multi-language translation capabilities, supporting text translation across dozens of languages. It’s designed for scenarios that require quick and accurate translations, such as content localization and multilingual customer support.Language Understanding (LUIS): LUIS is a machine learning-based service used to build natural language understanding into apps, bots, and Internet of Things (IoT) devices. It allows developers to define custom intents (representations of actions that a user wants to perform) and entities (parameters required to execute the action) relevant to their application’s domain and provides models that can understand user inputs in natural language.

Note

Microsoft has announced that LUIS will be retired on October 1, 2025. Microsoft recommends that organizations begin migrating to Conversational Language Understanding.

Conversational Language Understanding: This service provides tools to build conversational AI applications that can understand and respond to user queries in a natural way. It helps in developing sophisticated chatbots and virtual assistants that can engage with users conversationally.QnA Maker: This feature enables the creation of a conversational question-and-answer layer over your data, making it easy to build and maintain knowledge bases from your content, such as websites, documents, and FAQs.Custom text: This allows for the creation of customized NLP models tailored to specific industries or business needs. You can build custom classification, entity recognition, and single/multi-label classification models based on your unique datasets.Decision AI: Though not strictly limited to language processing, this feature integrates with the language service to aid in making informed decisions based on the text analysis, enhancing the decision-making processes within applications.

Azure AI Language was previously known as Text Analytics.

Identify Knowledge Mining workloads

You can think of knowledge mining a bit like an AI-powered search engine. Traditional search engines easily catalog and index traditional text pages. However, they may be less capable of ingesting unstructured data, such as directories of documents, spreadsheets, presentations, and images, and answering search queries for it.

Azure’s AI Knowledge Mining is a comprehensive suite of tools and services designed to extract valuable insights and knowledge from large volumes of unstructured data, such as documents, images, videos, and audio files. This suite leverages advanced artificial intelligence technologies, including natural language processing, computer vision, and machine learning, to enable organizations to understand their volumes of data.

Let’s look at the key feature of Azure’s AI Knowledge Mining:

Powerful text analytics capabilities