A machine learning algorithm is the method by which the AI system conducts its task, generally predicting output values from given input data. The two main processes of machine learning algorithms are classification and regression.
Machine learning (ML) algorithms are broadly categorized as either supervised or unsupervised. Supervised learning algorithms have both input data and desired output data provided for them through labeling, while unsupervised algorithms work with data that is neither classified nor labeled. An unsupervised algorithm might, for example, group unsorted data according to similarities and differences.
However, many ML approaches, including transfer learning and active learning, involve what are more accurately described as semi-supervised algorithms. Transfer learning uses knowledge gained from completing one task to help solve a different but related problem, while active learning allows an algorithm to query the user or some other source for more information. Both systems are commonly used in situations where labeled data is scant.
Reinforcement learning, sometimes considered a fourth category, is based on rewarding desired behaviors and/or punishing undesired ones to direct unsupervised machine learning through rewards and penalties.
Watch a tutorial about machine learning algorithms: