Supervised and Unsupervised learning in machine learning

Machine learning is a subject of artificial intelligence, which means programming computers in terms of algorithm so that they can learn from data. Supervised learning and unsupervised learning are two types of Machine Learning systems. One can easily tell the difference between them by the criterion: whether or not they are trained with human supervision.

Supervised learning and unsupervised learning are two extreme. The former completely depends on human supervision, while the latter doesn't need human supervision at all.  

Supervised learning

In supervised learning, you use the completely labeled training set to feed the algorithm. In other words, you label your desired solutions in the training set before feeding it to the machine.

There are two typical types of supervised learning:

• Classification. For example, the spam filter is trained with many example emails which are labeled as spam or ham. It must learn from this labeled training set to classify new emails.
• Regression. The task of regression is to predict a specific numeric value, such as the price of a book, given a set of features (publisher, age, cover, etc.) called predictors. To train the system, you need to feed it with many examples of books, including both their predictors and their prices.

The following are some of the most important supervised learning algorithms:

• k-Nearest Neighbors
• Linear Regression
• Logistic Regression
• Support Vector Machines (SVMs)
• Decision Trees and Random Forests
• Neural networks

Unsupervised learning

In unsupervised learning, you don't need to label the training set anymore. The machine will try to learn by itself without a teacher.

The following are some of the most important unsupervised learning algorithms:

• Clustering
• Anomaly detection and novelty detection
• Visualization and dimensionality reduction
• Association rule learning

For example, if you have a lot of data of your website's visitors which is a unlabeled training set. Then you could run a clustering algorithm to group different types of visitors. It will find connections between those visitors without your help.