Lesson video

Hello, my name is Mrs. Holborow.

Welcome to computing.

I'm so pleased that you've decided to join me for the lesson today.

In today's lesson, you'll be exploring machine learning.

What are the common uses of machine learning, and what is meant by the data-driven approach? Welcome to today's lesson from the unit "Data Science, AI, and Machine Learning." This lesson is called "Machine Learning," and by the end of today's lesson, you'll be able to recognise some of the applications of machine learning and some common machine learning models.

Shall we make a start? We will be exploring these keywords throughout today's lesson.

Let's take a look at them together now.

Machine learning.

Machine learning: an approach used to design and build artificial intelligence systems that is said to learn by using examples in the form of data.

Rule-based.

Rule-based: a way of designing systems using a set of predefined rules.

Data-driven.

Data-driven: a way of designing systems using data instead of step-by-step instructions or rules.

Look out for these keywords throughout today's lesson.

Today's lesson is broken down into two sections.

We'll start by identifying uses of machine learning, and then we'll move on to explain the data-driven approach.

Let's make a start by identifying uses of machine learning.

Artificial intelligence, AI, and machine learning, ML, systems are built from ideas and tools across many fields.

The diagram shows how these subjects overlap and connect.

So you can see we have a large field of computer science, and then we have AI that's sitting inside that.

And then we have a bit of an overlap between statistics, machine learning, and data science.

Machine learning is one approach used to design and build artificial intelligence systems. ML enables computer systems to learn from data and improve their performance on tasks instead of executing step-by-step instructions.

Okay, time to check your understanding.

I have a true or false statement for you.

Machine learning is the same as artificial intelligence.

Is this true or false? Pause the video whilst you have a think.

That's right, it's false.

Machine learning is one of the techniques that people can use to build artificial intelligence system models.

It's a subset of artificial intelligence.

Lucas says, "I'm not sure what the difference between AI and machine learning is." Do you know the differences? Maybe pause the video whilst you have a think.

Artificial intelligence is a term used to describe various techniques of using computer systems to help solve complex problems. Machine learning is a way of building data-driven AI applications.

The AI model is trained to make predictions.

The better the training data, the better the predictions.

Lucas says, "I make predictions in some of my lessons, like science experiments.

Is this the same?" Do you think it's the same? Maybe pause the video whilst you have a think.

The predictions made by machine learning applications are guesses based on data.

They cannot tell you exactly the right answer, but they can use data to make the best prediction.

So they're probably similar to the kind of predictions you make in other subjects, like science experiments.

What is the main goal of machine learning applications? Is it A, to automate manual tasks; B, to learn from data; or C, to provide exact answers to questions? Pause the video whilst you think about your answer.

Did you select B? Well done.

Remember, machine learning is used to learn patterns from data.

Izzy's got a question: "What are machine learning applications used for?" Common uses of machine learning applications include: recommendation systems, for example, music or TV programmes, image recognition, weather forecasting, and traffic and navigation systems. Let's now have a look at each of these in a bit more detail.

Recommendation systems. TV recommendation systems use large amounts of data, such as what people watch and how long they watch it for, to make recommendations about what users may want to watch next.

Machine learning systems identify and compare patterns in the data.

Image recognition or facial recognition.

Facial recognition is used to unlock devices such as smartphones or provide access to online accounts.

The system uses large amounts of data, such as lots of images of faces, to identify what makes faces unique.

Software identifies and compares patterns in the data.

Weather forecasting.

Machine learning applications are used to make predictions about the weather.

The system uses large amounts of data collected by weather stations and satellites to identify patterns in the data.

These patterns and trends can be used to make predictions about future weather.

Traffic and navigation systems. Online maps and navigation systems use machine learning to predict traffic, plan the most efficient routes, and estimate arrival times.

The system uses large amounts of data to identify patterns and to predict which routes may be busy and at what times.

Okay, we are moving on to our first task of today's lesson, task A, where we're going to identify uses of machine learning.

In your own words, describe two or three common uses of machine learning.

Pause the video whilst you have a go at the task.

How did you get on? Did you manage to describe two or three common uses of machine learning? Well done.

Let's have a look at a sample answer together.

Social media feeds use machine learning to make predictions about what you may want to see on your social media feed.

This uses large amounts of data, such as what videos you've watched previously, how long you've looked at images, and what posts you have commented on, to make predictions about your preferences and interests.

Image classification systems use machine learning to analyse and categorise items in an image.

A photo organisation tool on your smartphone gives you the ability to search your library by specifying the name of an object.

The application uses machine learning to identify photos that have a high probability of containing the object you are looking for.

Did you have some similar uses? Remember, if you used different examples in your answer, that's absolutely fine, but you can also pause the video here if you want to make any amendments to your answer.

So we've identified uses of machine learning.

Let's move on to explain what we mean by the data-driven approach.

Rule-based is a way of designing systems using a set of predefined rules, and you've probably come across a lot of rule-based algorithms in the past.

For example, a noughts and crosses programme is designed using rules of what moves to make in order to try and win the game.

The rules are defined by humans who are usually experts in the domain of the problem being solved, and the rules do not change.

AI systems built using a rule-based approach are also known as "good old-fashioned AI." Data-driven is a way of designing systems using data instead of step-by-step instructions.

Data-driven systems are suitable for solving problems where rules that cover every situation are difficult to produce or where the rules may change over time.

Current AI systems are mainly data-driven.

These systems are provided with lots of data.

An algorithm identifies patterns in this data to make a machine learning model.

This model can then make predictions on new, unseen data.

An example of this would be identifying animals in the wild.

As images of the animals could be taken in many different lighting conditions, at varying distances, and with animals in different positions, it would be very difficult to write an algorithm to do this using logic.

So instead, we could use a data-driven approach where we provide a model with lots and lots of different examples of animals in the world, lots of different photographs taken in different situations, and we could train the model to try and recognise particular animals.

So here we have a difference.

We have rule-based on the left-hand side, where we've got a set of predefined rules which are going to be followed by the algorithm.

And then on the right-hand side, we have the data-driven approach, where we have vast quantities of data which are provided and used to train the model.

Okay, I have a quick activity to check your understanding.

State which approach is used for each activity.

So the two approaches are rule-based and data-driven.

And the activities are: a spreadsheet that uses formulas to work out the total cost of a holiday, and online streaming service that recommends films based on a user's viewing habits.

Pause the video whilst you have a go.

Did you spot it? Well done.

A spreadsheet that uses formulas to work out the total cost for a holiday is rule-based, whereas an online streaming service that recommends films based on a user's viewing habits is data-driven.

A model is a representation of a real-world context.

A data-driven model is used to solve a problem.

Usually, they rely on a massive number of examples, lots of data, to detect patterns to create the representation.

So here we have an example of an AI chatbot.

It uses vast amounts of data to create a conversational model.

An application can use this model to hold a conversation.

Okay, time to check your understanding.

Have a true or false statement for you this time.

A fraud detection system that triggers a warning when someone spends more than 5,000 pounds in one day is an example of a data-driven approach.

Is this true or false? Pause the video whilst you have a think.

Did you select false? Well done.

This system has a predefined rule: the spend must be greater than 5,000 pounds a day.

So this is an example of a rule-based approach.

Okay, we are moving on to our second task of today's lesson, task B.

For part one, in your own words, explain what is meant by the data-driven approach.

And then for part two, give one example of a machine learning application that uses the data-driven approach.

Pause the video here whilst you have a go at the task.

How did you get on with the task? Did you manage to explain the data-driven approach? Well done.

Let's have a look at a sample answer together.

For part one, you were asked, in your own words, to explain what is meant by the data-driven approach.

A data-driven approach in AI means that instead of telling a computer exactly what to do, we provide it with lots of data, and it uses this data to detect patterns and make predictions.

The data-driven approach means letting AI systems learn from real-world examples.

For part two, you were asked to give one example of a machine learning application that uses the data-driven approach.

Media streaming apps that suggest movies, products, or music based on what they predict you'll like use a data-driven approach.

These systems track what users do, for example, what you watch, listen to, or search for.

They also compare data from different users.

All this data is then used to make suggestions and recommendations about what you should watch or listen to next.

Did you use this example, or did you have a different example? Remember, if you'd like to pause the video here and add any extra detail or corrections to your answers, you can do that now.

Okay, we've come to the end of today's lesson, "Machine Learning," and you've done a fantastic job, so well done.

Let's summarise what we have learned together in this lesson.

Machine learning, or ML, is a form of artificial intelligence that allows systems to learn from data and improve performance without being explicitly programmed.

Unlike traditional programming, where rules are manually written, ML models learn rules from the data itself.

As new data is collected, ML models can be retrained to become more accurate and relevant over time.

I hope you've enjoyed today's lesson, and I hope you'll join me again soon.

Bye.