Explaining Machine Learning to your Grandparent's

7 min readJul 6, 2020

Hello everyone, I am happy to have you here today to talk about a very interesting topic such as ML, but more precisely I am going to explain this subject in such a way that even a person that doesn't now anything about STEM (Science, technology, engineering, and mathematics) will understand it. So welcome!

Let’s start by understanding what is Machine learning (ML), Artificial intelligence (AI), Deep learning (DL) and the connections they have to each other

Artificial intelligence

As you can see in the Picture from above, AI is the biggest part of all of three, You can actually think of it as the Parent of Machine learning.

Serokel says: “AI is a science like mathematics or biology. It studies ways to build intelligent programs and machines that can creatively solve problems, which has always been considered a human prerogative”.

What a great definition of AI, to complement this, we can say that AI is now days everywhere… Like in hospitals where doctors can diagnosed patients, is in the internet when you get a recommendation of YouTube Videos to watch next, and also what the same company uses to control their content by removing videos classified as violent.

Forbes “The amazing ways Yotube Uses AI and ML”

Machine Learning

Before we explain that AI is the father of machine learning, so Machine learning will be the son! Yeah is pretty much like it, we can say that ML is the subset of AI and basically are systems that are trained rather than being program to solve problems. How? They use different algorithms like neural network to do it.

Machine Learning are Algorithms that gives computers the ability to learn from data and then make predictions and decisions — crash course

Algorithm

wow, but what an algorithm even means? Sounds like a complicated word isn’t, but it is just a fancy word to describe a set of instructions to perform a task. for example a recipe is the easiest example of it, because you fallow the instructions in a specific order. So for computers to work, they need to know what exactly to do. That it is why this word is very common use in computers field.

Now that we understand algorithms let’s keep learning about ML.

Arthur Samuel (1959) “Machine learning: Field of study that gives computers the ability to learn without being explicitly programmed” — Stanford University

So how computers get the ability to learn if we say before they have to follow an algorithm in order to operate? Well that is the main idea, combining this two concepts (algorithm and learning) is what actually Machine learning is about.

Let’s clarify that.

Computers at least not for now can’t learn as human do, instead they use Machine learning to predict results based on data. and repeating over and over the task, the easier it is to predict the result, focusing on teaching computers how to learn without the need to be programmed for specific tasks

There are three main components to teach a machine:

Data: the information we want the machine to analyze and learn. The samples can include numbers, images, texts or any other kind of data. It usually takes a lot of time and effort to create a good dataset.

2. Features: The variables or parameters the machine will look at. They are values that usefully characterize the things we wish to classify

3. Algorithms: The way the machine will solve the problem. Depending on the algorithm the speed and the accuracy will vary in each result.

example of training a machine with supervise learning

Machine learning algorithms or classical Machine learning

Supervised Learning
Unsupervised learning

Machine learning Algorithms using examples

Supervised Learning:

Is when the machine have a supervisor, or a mentor were the machine can learn from and get the right answers.

For example: “like whether it’s a cat in the picture or a dog. The teacher has already divided (labeled) the data into cats and dogs, and the machine is using these examples to learn. One by one. Dog by cat.” —

As we saw in the picture there are mainly two types of supervised learning algorithms

Regression: Its perfect when something depends on time. What actually the machine does is draw a line where the average of data is with mathematical accuracy.

Classification: We can be very familiar with this algorithm, which we probably use it all the time. An easy example could be classifying an object by its attributes, let’s say a pair of sucks organize by its color or size.

Unsupervised learning

It is completely oposite to the supervised learning. In this case the machine is left in its own and the data is not even sorted or classify. Of course there is not any supervisor, and basically the machine have to figure out the answers and the patters in its own.

Types of unsupervised algorithms

Clustering: The main goal is to select data with similar characteristics in some sense and making groups that relate to each other.

Association: Looks for patterns and discover interesting relation ship between the variables of databases. For example people who buy a new house tend to buy new furniture as well.

Others types of ML algorithms

Reinforcement learning: Is used to resolve problems when this one is not related to data at all. Let’s give an example of Reinforcement, creating a virtual city based on a real map for self-driving car so it can learn everything first here. It will learn how to drive around people and get confident in the virtual roads, once the robot pass the test it can actually be tested in real street’s. That is what is happening right now training auto pilots

Semi-supervised learning:

It is a combination of supervised and unsupervised learning, a mix of data sorted and labeled. “The programmer has in mind a desired prediction outcome but the model must find patterns to structure the data and make predictions itself.” — Yulia Gavrilova

Finally we got to the topic of Deep learning which is a subset of machine Learning.

Neural Networks and Deep learning

The field of deep learning is primary concerned with how to build computers systems that are able to successfully solve tasks requiring intelligence while the field of computational neuroscience is primarily concerned with building models of how the brain actually works

The main reason for not taking into account neuroscience in deep learning research today is that is very difficult to have information about how the brain works. Being able to monitor the activity of thousands of interconnected neuronas to understand the algorithms use by the brain is something we are not able to do at least now.

Even though we don’t know the algorithm our brains uses, we are inspired by the structure of it. Let’s see the approach of neural Network:

“Any neural network is basically a collection of neurons and connections between them. Neuron is a function with a bunch of inputs and one output. Its task is to take all numbers from its input, perform a function on them and send the result to the output.” — machine learning

Deep learning algorithms uses then a very complex neural networks.

Watch this video to find out more about Neural Networks

Now that we understand what is machine Learning let’s see which programming languages can we use in order to apply this field.

Python is a great option for machine learning at all levels, the simple syntax allows you to write and debug code easily

R This programming language appear in 1995, was written in C and FORTRAN. Is Ideal for initial purposes. was Initially use for scientific research. This language has a huge range of applications for collecting statistical data and data visualization.

Summary

In the above article, we learned about artificial Intelligence, Machine Learning its components and the various algorithms that are used for machine learning classification. We cover supervise and Unsupervised algorithms that are used for solving a variety of tasks. and finally we cover the subset of ML deep learning and neural networks.

Thanks for reading this far, I hope you enjoyed and learned something new. See you next time