Hi my friends, this is the article 4 / 1000!! For those who don’t know who I am, my name is Selim Chehimi and I’m an engineering student. I’ve been programming for 7 years now so you can tell that I really enjoy that 🙂. Those 7 years of programming lead me to Artificial Intelligence. I really want to be involved in AI so that’s why I’m sharing this article with you. I hope you’ll enjoy reading it as much as I’ve enjoyed writing it.
Today, I’ll look at the marvelous world of Machine Learning. When I first started to learn about Machine Learning, I must say that I understood nothing! It sounded so strange and mysterious and I think that’s the beauty of it.
What is Machine Learning? Here is the definition of Arthur Samuel in 1959: “Machine learning is the field of study that gives computers the ability to learn without being explicitly programmed”. Simple, right? Basically, you can consider the machine like a human infant. You’ll agree than an infant doesn’t really know or comprehend anything. However, the brain is looking for patterns to make sense of the world. It’s beautiful because it’s the same thing with machine learning! I want you to understand that the goal of AI is to create a machine that can mimic a human mind. That’s why it needs learning capabilities.
Here is a more recent definition given by Tom Mitchell, a professor at Carnegie Mellon University, in 1997: “If a computer program can improve how it performs a certain task based on past experiences, you can say it has learned”. So now we know that learning means improving.
Nowadays, we classify Machine Learning into 3 categories:
- Supervised Learning
- Unsupervised Learning
- Reinforcement learning
Follow me, I’ll explain you each one of these!
Let’s suppose that the goal of your AI is to recognize a cat from a dog. The first step to achieve this goal is to have a database with millions of cat and dogs. Then, we’ll regroup all cat images in the “cat category” and all dog images in the “dog category”. Finally, you can start to train (teach) your machine to recognize a cat from a dog! In supervised learning, all images in the database are tagged (cat or dog).
Again, let’s assume that the goal of your AI is to recognize a cat from a dog. Same thing, you have your database with millions of cat and dogs. However, there is no category: the AI will have to learn itself to find patterns to create those categories. I enjoy Unsupervised Learning because our AI will be like an autodidact: it’s like a reverse engineering.
Reinforcement Learning is like unsupervised learning, but the outcome is granted! Hmm, not clear? I’ll explain you with the Chess example. Let’s imagine that your AI is learning who to play chess. It will play thousands of games and each outcome will be granted. In other words, we’ll tell the AI if it wins the game or not. At the end, our AI will form a winning strategy. Impressive, right?
By the way, the greater the database, the more the machine can learn. Also, when our AI is well trained, we’ll give new unseen data. The algorithm will then use the past experiences to give you an outcome (cat or dog).
Now it’s time for me to show you ImageNet, one of the biggest image databases on the Internet. Here is how it looks when I search the word “cat”:
Every single image is labeled! It’s very useful for Machine Learning experiences. When I’m writing this, there are 14,197,122 images. Impressive, right? In Machine Learning, the most difficult is to have the dataset. You can imagine that it’s hard to build a dataset like ImageNet.
You know what is beautiful with machine learning? It’s everywhere! For example, your Facebook News Feed uses machine learning to personalize each member’s feed. If you liked one of your friends’ posts on Facebook, the algorithm will show you more of that friend’s activity earlier in the feed. The AI is simply using statistical analysis to identify model in the user’s data and populate the News Feed.
It’s the end of the article 4, I hope you liked it! Don’t hesitate to comment your thoughts in the comment down below and be sure to share this article!