The Beginner’s Guide To Machine Learning

In Artificial Intelligence by Andric Pasillas R.Leave a Comment

Are you tired of reading about the following words and not knowing what they’re referring to?

Machine learning.

You’ve come across this term many times while you were reading articles, and even your coworkers have mentioned it more than once.

But you have no idea what they’re talking about. Does it ring a bell? Well, buckle up, because things are about to change…

But first, a word of reassurance: unless you’re an artificial intelligence (AI) expert, AI terminology will probably sound peculiar and unfamiliar at first.

AI is also relatively new, so you’re not alone if you’re a little confused. However, once you’re acquainted with some of its basic notions and views, we promise it won’t seem so terrifying.

If you want to know the reasons behind all the fuss, and what machine learning is, but you don’t even know where to begin, this article is for you.

We know that learning a topic that’s so technical from scratch can be really frustrating. That’s why we decided to write this article—to introduce you to the subject the right way.

We really hope this post piques your curiosity and you decide to dig a little deeper on your own…

Because one thing is for sure: this fascinating field is about to change the way we live.

The Origins of Machine Learning

Machine learning was an ambitious idea born out of AI during the sixties.

Well, to be more specific, it was a subdivision of AI, resulting of the combination of computer sciences and neurosciences.

What this branch intended to study was pattern recognition (in engineering, mathematics, and computer science processes) and computer learning.

In the early days of AI, researchers were eager to find a way to make computers learn based only on data.


But as years went by, machine learning researchers started focusing on different issues—like probabilistic reasoning, statistically based research, information recovery and—of course—pattern recognition.

(All of these were also applied to engineering, mathematics, computer science and other fields related to physical or abstract objects).

This eventually caused machine learning to split from AI in the nineties. It became a discipline all of its own, though many purists refuse to accept that and still consider it a branch of AI.

Now, the main purpose of machine learning is to address and solve practical problems. This is where many of the numeric fields we mentioned before can be applied.

What Is It, Exactly?

As Arthur Samuel wrote in 1959, it gives computers the ability to learn without being explicitly programmed.

This definition is a little  obscure, so let’s rephrase it: AI uses algorithms  to give you insights and relevant conclusions from data, without the need for human intervention.

“Ok,” you say, “But what’s an algorithm?”

An algorithm is nothing more than a sequence or set of instructions that represent the solution to a determined problem.

So then, the purpose of machine learning is that people and machines work hand in hand.

Because machines are capable of learning just like a human being.

This is precisely what algorithms do; they allow machines to execute tasks, from the simplest to the more complex.


Initially, its functions were pretty basic and straightforward, like filtering emails.

Today, it’s a whole different story.

Now these technologies can make predictions about busy traffic intersections, detect cancer, make construction projects through advanced mapping in real time, and even tell us if two people are compatible.

How Does It Work?

Our main purpose  as learners is to develop the ability to generalize and associate.


But when it comes to a machine or computer, learning also means being able to perform familiar as well as unexpected or new tasks accurately and precisely.

And how is this possible?

By making a machine or computer replicate the cognitive faculties of a human being, creating models that “generalize” the information given to them, they are able to make predictions.

And the secret ingredient for all of this to happen? Data.

Actually, the origin and format of the data is not so relevant, since machine learning can process so many. All this data is known as Big Data.

What’s fascinating is that the machine or computer doesn’t even really perceive them as data, just a huge catalogue of useful examples it can use.

We could say the machine or computer algorithms fall into three main categories (even though there are many more): supervised learning, unsupervised learning, and reinforcement learning.

We will explain their differences below.

Types of Machine Learning


  • Supervised Learning

Supervised learning depends on previously labeled data.

A clear example of this would be a computer identifying the image of a car and telling it apart from the image of an airplane.

In order for this to be possible, a human must insert labels or tags to ensure the effectiveness and quality of the data.

In other words, these are problems we’ve already solved, but will keep coming up in the future.

The idea is that computers learn from a myriad of examples and, from then on, can do the rest of the math, so we don’t have to input any new information.

Examples: speech recognition, spam detection, handwriting recognition, and so on..

  • Unsupervised Learning

In this type of learning, the algorithm is deprived of its labels, so it doesn’t have any previous instructions to rely on.

Instead, it is provided with an enormous amount of data about an object’s characteristics (such as the  dimensions or features of an airplane or a car, for example). So the machine can determine what the object is, based on the information it has gathered.

Examples: identifying morphology of sentences, classifying information, etc.

  • Reinforcement Learning

Reinforcement learning involves learning through reinforcement.

To do this, the machine learns about a number of different situations through trial and error.

Even though the machine knows the results from the beginning, it doesn’t know what the best way to obtain them is.

So the algorithm gradually starts working out which patterns are successful, so it can then repeat them over and over until they are foolproof.

Examples: self-driving vehicles, decision-making, etc.

There are other more complex AI approaches for more specific tasks, but we won’t delve into those for the time being.

If you’re interested in learning more, you can check out some of the more advanced concepts right here. 

Machine Learning’s Current Applications And Overviews

Once you’ve applied machine learning techniques to solve problems that you thought were impossible to fix, it makes you realize that these technologies could solve virtually any problem—as long as there is enough data.*

*This is only true if the problem at hand is actually solvable.

For the modern consumer, machine learning is a key enabler that can help with lots of daily tasks. From translating services and weather forecasting, to guessing what you want based on your recent activities.

Machine learning’s simplicity and ease of use is unrivaled.

When it comes to the business world, many companies have decided to incorporate these kinds of technologies to their operating systems, with high hopes for improving and automating their routine processes.


According to the 2017 Global Digital IQ survey, 54% of the organizations interviewed are making significant investments in AI and that number is expected to increase to 63% in as little as three years.

A HubSpot survey on artificial intelligence also showed that 63% of people use artificial intelligence technologies on a daily basis, without even knowing it.

You may be wondering about the different ways machine learning can be applied.

Well machine learning is a system that processes and analyses data to obtain insights, so it can be applied within any field with databases large enough to do that.

For example, right now, it’s being used in/for:

  • Financial predictions and stock market fluctuations
  • 3D mapping and modeling
  • Fraud detection
  • Medical diagnosis
  • Search engine optimization
  • Voice recognition systems
  • Implementation of digital advertising campaigns
  • Classifying DNA sequences

One example of the last point is Adext AI, the first and only audience management platform that applies Artificial Intelligence and Machine Learning to digital advertising. The tool finds the best audience or demographic group for any ad and automatically manages budgets across several audience variations on Google AdWords and Facebook Ads.

It also guarantees a better cost per conversion (lower cost per sale or cost per lead) for all its Adext Partner agency’s accounts and campaigns, or the service will be FREE, and no management or optimization fees will be charged. And, as it currently amounts to an average improvement of +83% more conversions within its fist 10 day period, the only opportunity cost would come from not trying it.

To learn more about how exactly does Adext AI work, read this guide.

The Future of Machine Learning

We’ve seen how artificial intelligence could enhance our daily lives as consumers, but what about the business world?

Well, the conversations and comments from an endless stream of digital consumers (who increase in number every day) give these technologies an overwhelming amount of information daily.

The technologies are continuously obtaining new insights and knowledge, and spot trends faster than any human could.

While it’s true that Big Data is improving machine learning tremendously, making it much more efficient, it will require a great deal of human talent to make it flawless.

This is because computers are not skillful enough to determine the context of a situation.

They haven’t mastered reasoning.

This means that if we want machine learning to evolve and advance, we need experts from every academic field to train the machines and deliberately incorporate them into the processes they want to refine.


Finally, as with all new technologies, businesses will need to begin by understanding the basic principles of this industrial science so they can benefit from it.

In the meantime, all eyes are on this AI technology, since it will supposedly turn the world upside down…

As you can imagine, this was just a brief introduction to the intricate world of machine learning.

At a time when new technologies emerge in the blink of an eye, it’s easy to get lost in the gigantic maze of information and rising concepts.

But we hope this guide has helped you understand at least some of it, so you won’t be caught by surprise next time someone brings up the subject.

If you want to know more about what artificial intelligence can do for your business, these articles where we cover the benefits of these technologies may be just for you:

This article was written with the collaboration of Gabriel Kent, CTO of Adext. 

Leave a Comment