Many of you may already have heard of machine learning, which is not surprising, given the meteoric technological advances the world is currently experiencing, particularly in the field of AI. And I think that in a few years' time, the term will be familiar even to kindergartners.
In this article, we're going to focus on what machine learning is, its applications and importance in today's world, as well as its various algorithms.
Ready to get started?
Let's go ...
1- Introduction to machine learning
Machine learning is part of something bigger called artificial intelligence.
Basically, machine learning is a subfield of artificial intelligence, and its main specificity is that it enables machines to perform tasks without being explicitly programmed.
How cool is that?
Thanks to machine learning, machines are now able to imitate the human brain. This means that machines are now able to learn new things on their own and improve their performance.
Importance of machine learning in the modern world
From machine learning to deep learning and now generative AI, we can see that this field is truly on an exponential rise. Today, major tech companies - and not just them - are integrating AI into almost all their products, with the main aim of making people's daily lives easier. Take ChatGPT, for example, which is capable of browsing the Internet and answering different types of questions, while also generating code, to name but a few.
This shows the extent to which AI is now fully integrated into our society and daily life.
2- Types of machine learning algorithms
Basically, we can divide machine learning algorithms into three categories, mainly according to their objectives.
These main categories are
- Supervised learning
- Unsupervised learning
- Reinforcement learning
a- Supervised learning
In supervised learning the dataset is labeled. In other words, for this type of problem, in our dataset we are given the true example (target prediction) and the goal is to train the machine learning model to determine the relationship between the input features and the target prediction.
List of some common algorithms :
- Logistic regression
- Naive Bayes
- Decision trees
- Linear regression
- Support vector machines (SVM)
There are two types of problem in supervised learning, depending on whether the target variable is categorical (classification) or continuous (regression).
b- Unsupervised learning
As you might guess, here in unsupervised learning our dataset is unlabeled. Machine learning algorithms therefore attempt to discover patterns in the data without any guidance or instruction.
List of some common algorithms:
- k-means clustering
- association rules
c- Reinforcement learning
Reinforcement learning is basically the science of decision making. It's about making the right or optimal decision and being rewarded for it, or, conversely, being punished for it. Unlike supervised learning, the data set is not labeled, so there is no teacher. And in the case where there is no training data set, learning takes place through experience.
List of common algorithms:
- Q learning
- Deep adversarial networks
Some applications
Today, machine learning has applications in a wide variety of fields, including healthcare and robotics.
In robotics, I recently discovered that the most advanced Humanoid, Ameca, can see itself in the mirror.