What is Machine Learning (ML)?

Definition: Machine Learning is the study of computer algorithms and programs that can improve automatically through experience and by the use of data.

Machine learning is seen as part of Artificial Intelligence (AI).

What is Artificial Intelligence (AI)?

Definition: AI is the capability of machines to learn/imitate human behavior and patterns. With the help of AI, Machines can analyze images, comprehend speech, videos, interact in natural ways and make predictions using data.

In simple terms, we are enabling machines to learn quickly, enhance, make decisions based on what they learn (based on data inputs).

There are three main parts that we need to focus on for Machine learning on Azure.

Model

1. The task of creating models for teaching your AI application is the essence of machine learning.
2. A model is a way that you define what you want your machine learning implementation to learn.
3. You give it a model, which is a set of rules, and the application then starts playing this model repeatedly with the data you've provided. Over time, usually, very fast, the model will find patterns in the data that follow the rules you have provided.

Knowledge Mining

1. Knowledge mining is when you ask Azure Search, a service from Azure, to find insights that are present in your data.
2. This could be relationships between files, data based on geography, images with the same person in them, and much more.

Built-In Apps

1. Ready-made built-in apps on Azure, such as Cognitive Services for recognizing speech and language or Bot Services to create an automated process to answer questions in an interactive way.
2. Bots are what you may have seen on websites as "Assistants, ready to answer your questions!" or something like that.

Azure Bot Service

• The Azure Bot Service is a PaaS (Platform as a Service), offering that you can use to build your own bot.
• You can build any kind of bot from a simple question and answer bot that has a predefined list of questions and answers, surprise, to a complex virtual assistant bot for your customers.

The Bot Service has some cool benefits and features, too.

• Code or Visual:
  • You can create a bot using either the visual composer or by programming the code yourself.
  • The visual composer is a no-code way of getting your bot into play quickly.
• Language:
  • Add natural language and speech easily to catch common inquiries from customers before they take up time from support staff.
• Integration:
  • Integration channels, such as Facebook Messenger, Microsoft Teams, Twilio, and many, many more.
• Branding:
  • Put your own touch on the bot using your brand and keep ownership of any data that goes into or comes out of the bot.
  • A bot is a quick way to add an extra layer of customer service to your business or product.

Azure Cognitive Services

• AI as a service on Azure comes in a few different forms.
• They're all part of Azure Cognitive Services, which is a collection of tools to use in your own applications.
  • Vision:
    • The Vision service, which provides information about visual content found in images and videos; it will identify visual objects, people, and concepts, as well as caption the media for you too.
  • Decision:
    • The Decision service can make informed decisions based on data available to an app.
    • This could include identifying potential offensive language, getting notified of anomalies in your IoT network, or leveraging data analytics from your business.
  • Speech:
    • The speech service takes any spoken audio stream and then converts it into the text as a transcript.
    • The service can also identify unique voices and even verify a speaker, based on the speech.

Azure Machine Learning Studio

• The main tool for using machine learning on Azure is the Machine Learning Studio.
• This is a visual tool that can help you manage all your machine learning needs.
• It supports all the various programming languages, data sets, models, projects, and services that are available in Azure for machine learning.
• It has pre-made modules that you can just plug into your project and use straightaway.
• Some of the use cases for Azure Machine Learning Studio could be a Twitter sentiment analysis of all your followers, grouping specific types of flowers in photos, or recommending movies for users based on data about them.

Machine Learning Service

• End-to-end service:
  • Just as important as the Machine Learning Studio is, so is the actual Azure Machine Learning Service.
  • This is the end-to-end service for using machine learning pretty much everywhere on Azure.
  • It's a service that expands all of Azure and helps you use machine learning in more and better ways.
• Tooling:
  • It's a range of tools to help you build machine learning into your products and Azure processes.
• Automation:
  • It's a range of tools to help you build machine learning into your products and Azure processes.
  • And you get automated machine learning, which means Azure will pick up trends and processes automatically for you and create machine learning models that you can use.