Microsoft Azure AI: A Beginner’s Guide: Explore Azure Applied AI Services, Azure Cognitive Services and Azure Machine Learning with Practical Illustrations

By Rekha Kodali, Sankara Narayanan Govindarajulu and Mohammed Athaulla

Microsoft Azure AI A Beginner's Guide explains the fundamentals of Azure AI and some more advanced topics. The sole objective of the book is to provide hands-on experience working with the various services, APIs, and tools available in the Azure AI Platform. This book begins by discussing the fundamentals of the Azure AI platform and the essential principles behind the Azure AI ecosystem and services. Readers will become familiar with the essential services, use cases, and examples provided by Azure AI Platform and Services, including Azure Cognitive Services, Azure Computer Vision, Azure Applied AI Services, and Azure Machine Learning.

The author focuses on teaching how to utilize Azure Cognitive services to construct intelligent apps, including Image Processing, Object Detection, Text Recognition, OCR, Spatial Analysis, and Face Recognition using Computer Vision. Readers can investigate Azure Applied AI Services, including Form Recognizer, Metrics Advisor, Cognitive Search, Immersive Reader, Video Analyzer, and Azure Bot Service. Bot Framework and the Bot Framework Emulator will be explored in further detail, and how they can be used in AI applications to improve their conversational user interfaces. With Azure Machine Learning Studio, you will also learn to incorporate machine learning into your enterprise-level applications.

• Language: English
• Publisher: BPB Online LLP
• Release date: Apr 21, 2022
• ISBN: 9789355510594

Book preview

CHAPTER 1

Azure AI Platform and Services

This chapter gives a basic introduction to the services/components provided by Azure to build AI based applications, Azure Cognitive Services, and Azure Computer Vision. It gives an overview of the available APIs and then gives an introduction to some sample use cases possible with Azure Computer Vision.

Structure

This chapter will cover the following topics:

Platform provided by Azure to build AI-based applications

Azure Cognitive Services – Core Aspects

Conversational AI

Azure Computer Vision

Use Cases – Enterprise scenarios

Setting up developer tools

Objective

After reading this chapter, the reader will be able to understand the Azure AI platform, the core concepts of Azure Cognitive Services, various services available within Azure Computer Vision, and specific use cases leveraging Azure Computer Vision.

Azure AI platform

Intelligent applications leverage Artificial Intelligence to deliver rich, adaptive, personalized, immersive, contextual experiences to users intelligently and are autonomous. Natural interfaces of text, speech, and vision unlocking new categories such as conversational commerce and AR/VR help in building immersive applications. Immersive applications incorporate intelligent features such as emotion and sentiment detection, vision and speech recognition, language understanding, knowledge, and search. Azure AI Platform provides the capability to build such applications very quickly. Azure AI Platform can be broadly classified into four Services/Components as depicted below:

Figure 1.1: Components provided by Azure to build AI-based applications

Azure Applied AI Services

Azure Applied AI Services are built leveraging Azure Cognitive Services. They help in tasks like boosting literacy in the classroom, diagnosing, monitoring anomalies in metrics, knowledge from mining documents, document understanding, transcription analysis, and more scenarios.

Figure 1.2: Azure Applied AI Services

Azure Form Recognizer: It helps in identifying and extracting text, key/value pairs, and table data from form documents. It helps in ingesting text from forms and outputs structured data that includes the relationships in the original file.

Azure Metrics Advisor: It helps in identifying and fixing problems leveraging a combination of near-real-time monitoring, adapting models to specific scenarios and helps with diagnostics and alerts.

Azure Cognitive Search: It leverages built-in AI capabilities that help in identifying and exploring relevant content.

Azure Immersive Reader: It is an inclusively designed tool that implements techniques to improve reading comprehension for language learners, new readers, and people with learning differences.

Azure Video Analyser for Media: It helps extract metadata such as spoken words, faces etc from video and audio files.

Azure Bot Service: It helps in developing conversational AI experiences.

Azure Cognitive Services: Developers can rapidly consume high-level finished services that accelerate the development of AI solutions. They help in composing intelligent applications, customized to the organization’s availability, security, and compliance requirements.

Azure Machine Learning: Consists of a set of comprehensive tools and frameworks that help build, deploy, and operationalize ML models. They provide an extensive set of supported tools and IDEs and Azure ML studio that help in easily building and fine-tuning ML models and deploying them.

AI Infrastructure: AI Services and tools are backed by providing access to large scale infrastructure. It provides hyper-clusters of thousands of state-of-the-art GPUs, a breadth of AI hardware that includes the most comprehensive set of GPUs, and an array of general-purpose CPUs. It provides the latest high-bandwidth networks inside of every server.

We will look at some of the different types of applications that can be built using the Azure AI platform.

AI Services

AI services help access high-quality speech, vision, decision-making, language, and AI models through simple REST API calls and also provide the ability to create our own machine learning models.

AI services offered can be broadly classified as follows:

Figure 1.3: AI services

The following is an overview of AI Services:

Trained Services (Prebuilt)

Trained Services are Microsoft Azure Cognitive Services that offer a set of machine learning-based Azure Rest APIs that can be easily integrated into applications to infuse intelligence.

Use cases for Trained/Prebuilt Services: The following are a few use cases for leveraging Cognitive/Trained services:

Figure 1.4: Trained/Prebuilt/Cognitive Services

HealthCare: We can leverage OCR and Read APIs and other custom algorithms/logic as appropriate to go through large volumes of text that includes references to general entities (e.g. people’s names) and domain-specific ones (e.g. drug and disease names) that need to be connected and related. Sometimes we also need to combine this with imagery that’s analyzed in well-known ways (e.g. OCR) as well as apply leading-edge methods (e.g. AI-assisted diagnostics)

Oil and Gas companies: Oil and Gas companies have teams of geologists and other specialists that need to understand seismic and geologic data. They often have a huge library of PDFs with pictures of samples over sample sheets full of handwritten field notes. They need to connect places, people (domain experts), events, and navigate all this information to make key decisions. We can leverage OCR and other custom algorithms/logic to analyse and make decisions.

Insurance: Fraud Risk Analysis can be done by identifying anomalies in transactions using Azure Machine Learning, Face Recognition can be used as an additional authentication mechanism.

Conversational AI

Conversational AI helps in creating AI-based conversational interfaces. Azure provides Bot service which includes Bot Builder SDK and tools for end-to-end bot development and Bot Connector service to connect to multiple channels. We can create bots in a number of languages.

A Bot is a web service that communicates using a conversational interface and leverages Bot Framework Service to exchange events and messages between the Bot and the channel.

Bot functionality can be extended by adding additional features like NLP etc. Microsoft Cognitive Services such as Language Understanding Service (LUIS), can be added to the Bot interactions to make the conversation more intuitive.

The following are some ways of extending Bot functionality:

Figure 1.5: Extending Bot Functionality

The following table shows some of the additional features that can be leveraged to extend Bot functionality.

Table 2.1: Leveraging Bot Functionality

Use cases for conversational AI: A few use cases where conversational AI can be leveraged are as follows:

Bots can help Insurers give their customers an easy way to look up the status of a claim and ask questions about benefits.

Bots can help providers triage patient issues with a symptom checker, help patients find care, and look up nearby doctors.

Bots can help end customers have intelligent conversations with banks /retail websites.

Bots can be built leveraging language understanding models, bing knowledge and content out of the box. They are capable of learning from previous interactions.

Custom services

Custom services help in building applications using custom-trained models. Product recommendation, fraud detection for transactions, sales prediction are some examples of applications that leverage custom services. Azure provides an easy to use platform for developing/building, training and deploying custom machine learning models using tools like Visual Studio Code, Jupyter Notebooks and open source tools like Tensorlfow, PyTorch, and so on. It is also easy to embed these as services in custom applications through simple REST API calls.

Use cases for custom services: Given below are a few use cases where custom services can be leveraged

Figure 1.6: Use cases for Custom Services

Provide product recommendations, personalized recommendations for medicines, apparel, products, and so