Introduction to Artificial Intelligence – From Machine Learning to Computer Vision

Artificial intelligence has the potential to impact nearly every area of ​​life. In this article, the first of a two-part series explaining the buzzworthy technology, we look at different areas of AI technology and what it enables.

When we think about artificial intelligence (AI), most of us waver between excitement and anxiety about its rise. And with AI, like anything else, it's the unknowns that make us nervous.

AI and generative AI are creating incredible opportunities to dramatically improve government productivity and efficiency, enabling us to do more, better, faster, and easier. These technologies will enable us to run virtual simulations before taking real action, prevent adverse events, prepare for changing situations, detect concerns earlier and more accurately, engage in more meaningful ways, and better manage resources.

So what is AI?

Artificial intelligence is the science of designing systems that support and accelerate human decisions and actions. These systems perform tasks that historically required human intelligence, but artificial Intelligence is important for a reason: a simulation of human intelligence is performed by machines that are programmed to learn and think. AI does not replace humans, but rather augments and accelerates human actions and methods, increasing overall efficiency and productivity.

When talking about different types of AI, we sometimes refer to them as “disciplines of AI.” Each discipline performs a different type of task. There are three traditional disciplines of AI used by governments: machine learning, computer vision, and natural language processing. These three AI disciplines are interrelated and often overlap, and advances in one often influence advances in the others.

Additionally, Generative AI (GenAI) is a subset of Deep Learning, which is a subset of Machine Learning. The three technologies within GenAI are Large Scale Language Models (known as LLMs), Synthetic Data, and Digital Twins.

For those of you who have been hearing about or using ChatGPT or Copilot, they are built on top of LLM.

Before we talk about generative AI, let’s discuss traditional AI technology and how it works.

Machine Learning

Machine learning systems learn from data, identify patterns, and make decisions with minimal human intervention.

You may have taken a computer class in the past where you wrote conditional statements, or If-Then statements.

For example, a real estate agent might say, “If your property is adjacent to a lake, we'll increase its value by 10%.”

But with machine learning, you don't need to write “if then” statements. Machine learning models learn from the data you feed them. The more data you feed the model, the more accurate it becomes.

Machines can ingest large amounts of data, extract key features, decide how to analyze it, write the code to perform that analysis, and generate intelligent output — all through an automated process.

For example, say a computer is assessing the value of a property. It considers thousands of properties. It compares properties that are adjacent to water with those that aren't. From the data it reads, the computer determines that properties that are adjacent to lakes are 11% more valuable than those that aren't. This rule is not a fixed rule. In fact, any changes to the data input into the system will cause the rules and outputs to change. Typically, the more data the system has to process, the more sophisticated the answer becomes.

Deep Learning

Deep learning is a subset of machine learning that teaches computers to process data in ways inspired by the human brain. Just as neurons in the brain send information between brain cells, layers of deep learning nodes work together to process data and solve problems. Deep learning can be likened to the process of teaching a child to recognize animals, through layers of learning, continuous testing and correction, and examples diverse enough to generalize to new situations. Like a child, deep learning gets better with practice and deepens its understanding with each new example. Deep learning is used for natural language processing, computer vision, and generative AI.

Natural Language Processing

Natural language processing enables understanding, dialogue, and communication between humans and machines.

NLP enables computers to read text, listen to and interpret speech, measure sentiment, and determine which parts are important. The overall goal is to take raw language input and use linguistics and algorithms to transform or enhance the text to provide greater value.

Natural language processing is closely related to text analysis, a machine learning technique that counts, group, and classify words to extract structure and meaning from large amounts of content.

All of these areas of AI contribute to each other. Computers can augment human efforts to analyze unstructured text with AI, using a combination of natural language processing, machine learning, and linguistic rules. NLP and text analytics are used together in many applications, including research discovery, subject matter expertise, and social media analysis.

For example, criminal investigations typically involve a huge number of intelligence reports. Not only are these reports extremely time-consuming to read, but the process of extracting key people, addresses, phone numbers, and relationships that are evidence relevant to the case is tedious. To obtain new information from a crime report, previously read reports must be combed through, making the process repetitive and time-consuming.

ML can be used to extract and tabulate people, places, events, objects, phone numbers, and email addresses from long stretches of text, such as crime reports, speeding up information discovery.

Applying linguistics and analytics, NLP systems can infer nuances, such as sentiment, from the sentences in a report by identifying syntax (the structure, arrangement and order of words and phrases) and semantics (the grammar of a sentence, the grammar of its parts, the grammar of its parts, the grammar of its parts). meaning Analysis of language with a focus on words, phrases, sentences and “discourse” How language is used in context to convey meaning.

Computer Vision

Computer vision is the branch of AI that trains computers to interpret and understand the visual world, enabling systems to visualize, identify, and process images and videos in a way that is similar to human vision.

Using deep learning algorithms, machines can accurately identify and classify objects in images and videos and react to what they “see.”

Applications of computer vision include facial recognition and surveillance image analysis.

This graphic shows how computer vision works.

On the left is a portrait of a famous American. The image is pixelated and a number is assigned to the shade of each pixel. On the right is how the computer defines the image.

There are many computer vision techniques that can be used to analyze images and videos, some of which are listed below:

• Image segmentation involves dividing an image into regions or parts and inspecting them separately.
• Object detection to identify specific objects in an image, or more advanced object detection to recognize many objects in a single image, such as the playing field, attacking players, defending players, the ball, etc. These models create a bounding box using X,Y coordinates and then identify everything within the box.
• Pattern detection is the process of recognizing repeating shapes, colors, and other visual indicators in an image.
• Edge detection is a technique used to identify the outer edges of objects or scenes in order to more accurately identify things in an image.
• Image classification to group images into different categories.
• Feature matching is a type of pattern detection that matches and classifies similar features in images.

This is the first of a two-part series looking at how AI and generative AI work, to help civil servants become familiar with the characteristics and capabilities of different AI technologies and understand the type of AI needed to tackle different tasks. Stay tuned for our next article on generative AI.

To learn more about how AI can benefit your government organization, contact Jennifer Robinson, Global Public Sector Strategy Advisor, SAS. [email protected] Or visit our website at sas.com/public-sector.