This type of machine learning involves using labeled data to train algorithms to make predictions or classifications. In this technique, the algorithm learns to map inputs to outputs by minimizing the difference between the predicted and actual output.