artificial intelligence - What is the difference between supervised learning and unsupervised learning? -

In case of artificial intelligence and machine learning, can you provide a basic, easy explanation with an example?

Since you ask this very basic question, it seems to be the only thing to know the machine worth a shot. The machine learning algorithm is a class that is data-driven, that is the data that "tells" is a "good answer", unlike "normal" algorithms. Example: To detect face in images, a hypothetical non-machine learning algorithm will try to determine what the face is (like a round disk with a dark area, where you expect eyes). A machine learning algorithm does not have such coded definition, but will be "learn-by-example": You will show many images of face and no face and a good algorithm will eventually learn and it will be able to predict whether an invisible image The face is.

This specific example of face detection is supervised , which means that your example should be labeled , or explicitly say who There are faces and people who are not.

Your examples in the Unsorganized algorithm are not labeled , that means you do not say anything. Of course, in such a case the algorithm itself can not face "invention", but it can try data in different groups, e.g. It can detect that the faces are very different from the landscape, which are very different from the horses.

In another answer it has been mentioned (though, incorrectly): there are "intermediate" forms of supervision, i.e. semi supervised and active learning Technically, these supervision methods have some "smart" ways to avoid a large number of labeled instances. In active learning, the algorithm itself decides what you should label (for example it can be very convincing about the scenario and the horse, but it can ask you to confirm whether the gorilla actually Is a picture of a face). In semi-supervised education, there are two different algorithms that begin with examples labeled, and then each other tells "something" that they think of a large number of unreserved data that they learn from this "discussion" Are there.