Chapter 01: Introduction to Artificial Intelligence in Azure
Introduction to Artificial Intelligence
The field of computer science related to the theory of technologies that “think like creatures” and perform tasks such as learning, planning, reasoning, problem-solving, and identifying patterns is known as Artificial Intelligence.
Artificial Intelligence (AI) is the study of automating intelligent practices at present only achievable by people. The power framework has developed enormously over the years and the evolution of AI has advanced the development of human society in our own time, with dramatic revolutions shaped by both principles and techniques. In this chapter, we study the evolution of AI from the beginning, then the principles, domains, and applications of AI, as well as conversational AI, and how it is rapidly becoming highly significant in the Microsoft Azure domain.
What is AI?
The essentials of Artificial Intelligence (AI) are very basic. AI is about encoding the world into a form of knowledge representation. What is the significance of it? It implies that the machine has an approach to storing and accessing information. At the point when machines do that, they will be able to reuse various components of the information as they see fit. The famous scientists Stuart Russel Peter Norvig state that Artificial Intelligence is a modern approach:
“AI is the study of agents that receive percepts from the environment and perform actions. Each such agent implements a function that maps percepts sequences to actions, and we cover different ways to represent these functions, such as reactive agents, real-time planners, and decision-theoretic systems.”
Constructing an AI system is the careful progression of reverse-engineering human traits and capabilities in a machine. To understand in what way Artificial Intelligence essentially works, one prerequisite is to deep dive into the different
sub-domains of AI and understand how individual domains could be applied to the numerous fields of the industry.
Machine Learning (ML) is often the foundation for an AI system that teaches a machine how to make inferences and predictions based on experience. It is an application of AI that identifies patterns and provides systems
with the ability to automatically learn and improve from experience without being explicitly programmed. It focuses on the development of computer programs that can access data and reach conclusions by evaluating the data.
It saves people and businesses time and helps them make better decisions.
In data analysis, the identification of exceptional items, events, or observations that raise reservations by differing considerably from the majority of the data is known as Anomaly Detection –
the ability to robotically detect errors or unusual activity in a system. It is all about finding patterns of interest (outliers, exceptions, peculiarities, etc.) that deviate from expected behavior within a dataset(s).
This is the capability of software to interpret a domain visually using cameras, video, and images. Computer vision algorithms attempt to understand an image by breaking it down and reviewing its different parts. This supports machine classification and the system learns from a set of images, helping make good output decisions based on previous observations.
Natural language processing
Natural Language Processing (NLP) is the capability of a computer to interpret spoken or written language and act in response. NLP is the science of reading, understanding, interpreting a language using a machine. Once a machine understands what the user intends to communicate, it responds accordingly.
This is the capability of a software “agent” to participate in a conversation. Conversational AI can handle requests at a greater volume than human beings, by providing relevant and correct
information faster and with greater accuracy and complexity over time. The latest level of conversational AI applications is Virtual Personal Assistants. Examples of these are Amazon’s Alexa, Apple’s Siri, and Google’s Home.
Understand machine learning
Machine Learning has quickly become one of the most critical domains in the field of computer science. It is a subset of AI techniques that uses statistical methods to enable machines to improve with experience. Nowadays, many people in the industry are taking advantage of this domain to solve problems by developing products that are based on machine learning.
Computational statistics is closely related to the subset of machine learning that emphasizes making predictions using workstations. It is an application of AI that provides systems with the ability to
robotically learn and improve from experience without being explicitly programmed.
Machine learning allows the analysis of huge quantities of statistics or data. However, it generally provides quicker, more precise outcomes, identifying cost-effective opportunities or
levels of risk; it may also need extra time and resources to train it appropriately.
Merging machine learning with AI and cognitive technologies can make AI even more operational in
processing large volumes of information
How machine learning works
The Machine Learning mechanism works in a similar way as human learning. For example, if a child is shown pictures of specific objects, the child learns to recognize and differentiate between them. Machine Learning
works in the same way: by taking data input and definite commands, the computer is permitted to “learn” to recognize certain objects (persons, objects, etc.) and to differentiate between them. For this purpose, the software
is provided with data and is trained. For example, the computer programmer can tell the system that a particular object is a human being (=”human”) and another object is not a human being (=”no human”). The software collects continuous
feedback from the computer programmer. These feedback indications are used by the algorithm to adapt and improve the model. The model is furthermore enhanced with each new data set loaded onto the system, so that in the end it can distinguish
among “humans” and “non-humans”.
There are three major parts in machine learning systems, as follows:
Model: the system that makes predictions or identifications
Parameters: the indicators or factors used by the model to form its decisions
Learner: the system that adjusts the parameters and in turn the model by looking at differences in predictions against the actual conclusion
Figure 1-01: The Working of Machine Learning
Machine learning algorithms build a model based on sample data, known as “teaching data“, to make predictions or decisions without being explicitly programmed to do so. Machine learning algorithms are used in a wide variety of applications, such as email filtering and computer vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks.
Machine learning in Microsoft Azure
Azure Machine Learning is a distinct and rationalized service that distributes a complete data science platform. It supports both code-first and low-code experiences. Azure Machine Learning Studio is a web portal in Azure Machine Learning that contains low-code and no-code options for project authoring and asset management.
- Azure Machine Learning can be used for any kind of machine learning, from classical ML to deep learning, supervised, and unsupervised learning.
- You can even use MLflow to track metrics and deploy models or Kubeflow to build end-to-end workflow pipelines.
Understand anomaly detection
Outlier detection is also known as Anomaly Detection. It is the identification of unexpected actions, observations, or items that differ considerably from the norm. It is any process that discovers the outliers of a dataset; those elements that are not appropriate. These anomalies might point to unfamiliar network traffic, uncover a sensor on the fritz, or simply detect data for cleaning before exploration.
Anomaly detection in Microsoft Azure
Anomaly detection, or outlier analysis, is a phase in data mining that categorizes data points, events, and observations by deviating from a dataset’s common behavior. Anomalous data can specify critical incidents, such as a technical glitch or potential opportunities, for example, a change in consumer behavior.
Anomaly detection abilities can be easily embedded into your apps so that users can quickly recognize glitches. By using an API, Anomaly Detector consumes time-series data of all types and picks the best-fitting detection model for your data to make sure of high accuracy. You can customize the service to identify any level of an anomaly and deploy it wherever it is most needed, from the cloud to the intelligent edge with containers. Azure is the only key cloud provider that deals with anomaly detection as an AI service.
Understand computer vision
Computer vision is the field of learning regarding what way computers understand and recognize digital images and videos. This interdisciplinary field simulates and automates these fundamentals of human vision systems using sensors, computers, and machine learning algorithms.
Computer vision has been given a particular focus. For example, agriculturalists are now capable of photographing specific crops and upload them to the Intello Labs database to learn about them. Depending on requirements, they can learn about weeds, pests, diseases, and more.
Computer Vision models and capabilities
Computer vision defines the process of using digital images and videos to gain stronger insights from the user end. Information from data sources is pulled out automatically, analyzed, and understood. In the current age, it has been used for medical image analysis, facial recognition, surveillance, pollution observation, and further highly innovative systems.
According to Prof. Fei-Fei Li, computer vision is well-defined as “a subset of mainstream artificial intelligence that deals with the science of making computers or machines visually enabled, i.e., they can analyze and understand an image.” Human vision starts at the biological camera’s “eyes” and
proceeds one picture about every 200 milliseconds; computer vision, surprisingly, achieves that input in a machine.
The following are significant illustrations of computer vision model applications that are makeover industries:
The Automotive Industry
In recent years, with the support of computer vision practices, the automotive industry has been dedicated to the development of self-driving cars. Autonomous cars should be capable of tracking all surrounding objects with cameras and responding according to whatever is happening around them.
AI-driven computer vision can be used to improve increase crop yields as it notifies farmers of effective growth approaches, crop quality and health,
and soil conditions. Image classification techniques are being used
to automate quality control of crops by grading and sorting them based on their physical attributes.
Computer vision has multiple applications in the field of healthcare. Computer vision methods can prove to be life-saving for numerous patients. They permit medical professionals to monitor conditions and diseases and make diagnoses that guide how doctors recommend medications and provide treatment. They can also identify fatal illnesses. These applications similarly improve medical processes as they moderate the time doctors use analyzing medical images, offering additional time for consultations.
Face recognition is one of the most common applications of computer vision. It is being used as authentication for security purposes by being integrated
into security cameras to harvest real-time information from video feeds.
Banks and other financial organizations
have now started to apply computer vision. Some organizations permit their customers to open accounts using facial recognition for verification. This methodology has been demonstrated to be less time-consuming than traditional
pen and paper approaches.
Good advertising can benefit consumers to distinguish between products and services from their visual properties; tracking, and by visualizing emotional reactions
Computer vision technologies can make significant insights into education, greatly improving teaching methods and personalized learning. AI power-driven cameras can assistance teachers, instructors, and educators, monitor their students’ progress to improve classroom interactions, and enhance the learning experience.
Computer vision services in Microsoft Azure
The following are the key benefits of computer vision services in Microsoft Azure:
Extract rich information from images and video
With the support of computer vision, you can boost content discoverability, automate text extraction, analyze video in real-time, and produce products that more people can use by embedding cloud vision abilities in your apps, part of Azure Cognitive Services. Use visual data processing to label content with objects and concepts, extract text, generate image descriptions, moderate content, and understand people’s movement in physical spaces.
In real-time, recognize how people move in space.
Run computer vision in the cloud or on the edge.
Deploy anywhere, from the cloud to the edge
Run computer vision in the cloud or on-premises with containers. Apply it to diverse scenarios, such as healthcare record image inspection, text extraction of protected documents, or study how people move through a store, where data security and low latency are dominant.
Understand natural language processing
Natural language processing (NLP) refers to AI employing a language such as English, Spanish, French, Urdu, etc. Language can be a theoretically motivated range of computational techniques for analyzing and representing present text/speech at one or more level of linguistic analysis in order to achieve human-like language processing for the expansion of tasks or applications.
Importance of NLP
- NLP helps us to make communication and handling between the system easy.
- It helps to have access to large social data available online.
- It improves the efficiency and accuracy of documentation from large databases.
Natural language processing in Microsoft Azure
Natural language processing is used for tasks such as sentiment analysis, topic detection, language detection, key phrase extraction, and document categorization.
Figure 1-02: Natural Language Processing in Microsoft Azure
Understand conversational AI
Conversational AI deals with technologies such as chatbots or voice assistants which users can inquire about. They use immense volumes of knowledge, machine learning, and tongue processing to assist imitate human interactions, recognize speech and text inputs, and decode their meanings across various languages.
Conversational AI in Microsoft Azure
Online chatbots and voice assistants for customer support services and omnichannel deployment frequently come to mind when considering conversational AI. Most conversational AI apps have extensive analytics built into the backend program, helping ensure human-like conversational experiences.
Experts currently consider conversational AI’s applications to be quite weak AI, as they’re focused on performing an extremely narrow field of tasks. Strong AI, which remains a theoretical concept, focuses on human-like consciousness that can solve various tasks and solve a broad range of problems.
Despite its narrow focus, conversational AI could be a particularly lucrative technology for enterprises, helping businesses become more profitable. While AI chatbots are the most familiar type of conversational AI, there are many other use cases across the enterprise. Some examples include:
Online customer support: Online chatbots are replacing human agents along the customer journey. The answers to >frequently asked questions (FAQs) on topics such as shipping, or providing
personalized advice, cross-selling products, or suggesting sizes for users, change the way we expect customer engagement across websites and social media platforms. Examples are comprised of texting bots on
e-commerce sites with virtual agents, messaging applications, for eg: Slack and Facebook Messenger, where tasks are mostly done by virtual assistants and voice assistants.
Companies can become more accessible by decreasing entry hurdles , peculiarly for users who use assistive technologies. Commonly used features of conversational AI for
these groups are text-to-speech dictation and language translation.
- HR processes Many human resource processes are often optimized by using conversational AI, for example, employee training, onboarding processes, and updating employee information.
Healthcare: Conversational AI can make healthcare services easier to access and reduce prices for patients, while also enhancing operational effectiveness, making the executive process, just like claim processing,
Internet of things (IoT) devices: Most households now have at least some IoT devices, from Alexa speakers to smartwatches to cell phones. These devices use automated speech recognition to tie in with end-users.
The more familiar applications include Amazon’s Alexa, Apple’s Siri, and Google’s Home.
- Computer software: Many tasks in an office environment are simplified by conversational AI, for example search autocomplete when you search on Google and spell-check.
While most AI chatbots and apps now have rudimentary problem-solving skills, they are enhancing the cost-effectivity on iterative customer support interactions, freeing up personnel assets to trust more engaging customer interactions.
Overall, conversational AI apps are being developed to replicate human conversational experiences well, leading to improved rates of customer satisfaction.
Artificial Intelligence systems can act unpredictably for a variety of reasons. These software tools can support you to understaned the behavior of your AI systems so that you can improve them and tailor them to your needs. Microsoft researchers are working with the broader academic community on the advancement of responsible AI practices and technologies. At Microsoft, AI software development is conducted following a set of six principles designed to ensure that AI applications deliver amazing solutions to difficult problems, without any unintended negative consequences.
AI systems must treat all people equally and fairly. For instance, assume that you have created a machine learning model that provides support to loan approval applications for a bank. The model needs to make predictions about whether a loan should be granted without incorporating any kind of bias, for example gender, ethnicity, or any other factors that might result in an unfair advantage or disadvantage to particular groups of applicants.
Azure Machine Learning includes the ability to interpret models and quantify the extent to which each feature of the data affects the model’s prediction. This ability helps data experts and inventors to recognize and mitigate bias in the model.
Reliability and safety
AI systems must accomplish reliability and safety. For instance, consider an AI-based software system for an autonomous automobile, or a machine learning model that diagnoses patient symptoms and endorses prescriptions. Unpredictability in these kinds of systems can result in extensive risk to human life. AI-based software application development must be subjected to rigorous testing and deployment management procedures to ensure that they work predictably before release.
Privacy and security
Privacy and security are key elements in AI systems. The machine learning models on which AI systems are based rely on large volumes of records, which may hold personal information that must be kept private. Even after models are trained and the system is in production, it incorporates new data to make predictions or take actions and this data may be subject to privacy or security concerns.
AI systems must engage people and empower everybody. AI should bring profits to all parts of society, regardless of physical ability, gender, sexual orientation, ethnicity, or other characteristics.
AI systems should be reasonable. Users should be fully prepared to be aware of the purpose of the system, how it works, and what restrictions may be probable.
For AI systems, people should be accountable. Designers and developers of AI-based solutions should work within a framework of governance and organizational principles that ensure that solutions meet well-defined ethical and legitimate standards.
Figure 1-03: Mind Map of AI on Azure
You are designing an AI system that empowers everyone, including people who have hearing, visual, and other impairments.
AI-900: Microsoft Azure AI Fundamentals
- Covers Complete & Official Exam Blueprint
- Summarized content
- Case study based approach
- Ready to practice labs
- Exam tips
- Mind maps
- 100% passing guarantee