# How We Learn — and How Machines Do It Too

Oct 23, 2018 · 5 min read

When you were born, you didn’t know anything.

You didn’t know where you were. You couldn’t count or talk. You didn’t know the alphabet. You didn’t even know what your name was. Then you opened your eyes and everything changed.

You began to see.

Sure, it was blurry and everything was black and white. All you could see were distinct blobs and shapes moving about. Still, with your newfound vision, you began perceiving the world. Eventually, you remembered faces and what your favourite toys looked like. You also could hear the sounds and smell the scents around you. You used your five senses gather information and learn about your surroundings.

When you were around 9 months old, you decided to embark on an adventure; you took your first steps. You took one or two, then fell right down. Maybe you tried again and stumbled again. It would take many more months until you could walk well. You practiced countless times to fully master walking.

You did hundreds or even thousands of addition and subtraction problems before it became second nature. It took practice to learn the basic mathematical operations until it became ingrained in your memory. It takes practice to learn and master skills.

You were a very curious baby. Everything amazed you and you would touch everything in sight. Remember when you touched the stove? It looked so fascinating! You couldn’t resist and reached out, only to scream out in pain. From experience, you learned to never touch the stove again.

In school, when you spoke to the class without raising your hand, you were reprimanded by your teacher. It took experience to learn to raise your hand when speaking. It takes experience to learn from mistakes.

As a baby, you probably tried dozens of puzzles, way more than you could ever count. In the beginning, you just tried to put any pieces together. But they rarely ever fit! You had to try a bunch of different combination of pieces to get it right. You used trial and error to learn to do puzzles and find a strategy that worked for you.

When you started algebra, you stared at the equations 5+x=7 and 6-y=2. You just guessed different possible values for each variable and eventually got x=2 and y=4. It took trial and error to figure out how algebra worked and develop ways to find solutions systematically. It takes trial and error to learn and develop strategies to solve problems.

This is how you learn. This is how you’ve figured out how to do everything. Your knowledge of the world helps you live your everyday life to the fullest. But this isn’t just you. It’s not even how every single human learns. These strategies are what EVERYTHING uses to learn, even machines.

Enter machine learning.

Machine learning is a huge field that is taking over the world. In simple terms, it’s telling the machine to perform a task without being explicitly programmed to do so. In other words, the machine learns to do something all by itself, without every process having to be coded. And it does so just exactly like you and me.

First, we input data into the machine learning algorithm. Data is just like the information we gain from our senses. Without data, the machine is a newborn baby; it knows nothing. The program uses trial and error to find a model that fits the data best (think of the model as the brain, it receives data and produces a result). It tries different values and uses a strategy to create the optimal model. To train the model well, we give it a lot of data, just like how we humans must practice many times in order to master something. During training, the machine may get things wrong. It compares its generated answers to the answers given in the data and changes its model accordingly, learning through experience.

This is the process of how a specific machine learning type, called supervised learning, works. As you can see, machines can learn on their own (as long as they are given data and have a machine learning algorithm), just like us! Some types of machine learning don’t even require data! This is why we say that machines can have artificial intelligence or AI.

Of course, we’ve only scratched the surface here. This is just a simplified version of how supervised machine learning works. It can implement different algorithms like gradient descent or the normal equation to solve linear or logistic regression problems and in order to minimize the cost function for supervised learning problems. There are also other types of machine learning, like unsupervised and reinforced machine learning. We’ll touch on that later in future articles.

The key takeaway here is that humans and machines can learn in the same way! The difference is that machines can learn much faster and more efficiently. They can run through their data and analyze them orders of magnitudes faster than the speeds of which humans can have experiences or practice learning skills. Machines never get tired or distracted and can learn continuously, unlike humans.

As you can see, machine learning really is the future! There’s so much it can help us do to improve our lives and make the world more efficient. Unfortunately, machine learning can also be used with malicious intent to harm others. With such a powerful tool in our hands, how will we use artificial intelligence? That’s for you to decide.

Thanks for reading! Be sure to follow me if you want to learn a lot more about machine learning and AI! :)

### By Data Driven Investor

In each issue we share the best stories from the Data-Driven Investor's expert community. Take a look

Written by

Written by