When you open your mailbox, you do not see that much of spam these days as they are automatically filtered in a separate folder. If you receive a document in an unknown language, you can easily translate it to any other language you understand in just one click. Your bank blocks your credit card based on some suspicious activity before you even realize that it was stolen. Your favourite streaming service always suggests the movies you are fond of, and the special offers you get from your favourite stores are always on point.
What do all of these, and many other technologies share in common, besides making our lives way easier? All of them are powered by Machine Learning (ML), a sub-field of Artificial Intelligence (AI), which is booming nowadays.
In 1959, Arthur Samuel, a computer scientist who pioneered the study of artificial intelligence, described machine learning as:
[Machine Learning is the] field of study that gives computers the ability to learn without being explicitly programmed.”
In other words, we should use Machine Learning when we want to let the compute do something without explicitly writing the code that explains how one should do it.
A more technical definition given by Tom M. Mitchell’s (1997) :
“A computer program is said to learn from experience $E$ with respect to some class of tasks $T$ and performance measure $P$, if its performance at tasks in $T$, as measured by $P$, improves with experience $E$.”
Let’s take a look at a simple example:
If you just download a copy of Wikipedia, your computer has a lot more data, but it is not suddenly better at any task. Thus, it is not Machine Learning.
Machine Learning is the study of Algorithms that use examples objects (data), to learn mappings from objects to desired outcomes.
Given an input, we will get a prediction based on some model. Some examples:
Machine Learning tries to identify regularities in the world, by learning from examples (“data”). These regularities should generalise: work beyond the specific examples the model has seen before. Machine Learning methods can predict future data, or perform other kinds of decision making under uncertainty (such as planning how to collect more data!).
Generalization means coming to general conclusions from (a limited number of) specific observations. Based on this prior knowledge, we want to predict for new, unseen data.
ML is everywhere in nowadays world. For example: