Supervised Learning vs. Unsupervised Learning

Machine Learning (ML) algorithms are often broadly divided into two groups: Supervised Learning Algorithms and Unsupervised Learning Algorithms.

As a reminder, Machine Learning is a subfield of Artificial Intelligence (AI) and describes algorithms that are able to improve themselves through experience.

Supervised ML algorithms are the most common form of ML and are used with the goal of making accurate predictions based on new data. To do this, the algorithms require "labeled" data sets, i.e., data sets in which the correct output is assigned to each input. Using this data set, the algorithm is able to learn a function that allows it to make predictions based on new data. An example of the use of supervised ML algorithms is spam detection for e-mails.

Unlike Supervised ML algorithms, Unsupervised ML algorithms do not require labeled data sets. This is because Unsupervised ML algorithms are used with the goal of finding patterns and structures in data and/or extracting information. Compared to Supervised ML algorithms, Unsupervised ML algorithms are usually much more complex. An example for the use of Unsupervised ML algorithms are recommendation systems as they are used in online stores.

Source (translated): IBM