A subfield of AI that develops algorithms allowing machines to learn patterns from data and make predictions or decisions — without being explicitly programmed for each scenario.
In Depth
Machine Learning is the engine behind most modern AI applications. Instead of a programmer writing explicit rules — for example, marking as spam any email containing the word lottery — an ML system is given thousands of examples of spam and legitimate email, and learns to distinguish them on its own. The system extracts statistical patterns from data and builds an internal model that generalizes to new, unseen examples.
ML can be broken into three main paradigms. In Supervised Learning, the model learns from labeled examples (input-output pairs). In Unsupervised Learning, the model finds hidden structure in unlabeled data — grouping similar customers together, for instance. In Reinforcement Learning, an agent learns by interacting with an environment and receiving rewards or penalties based on its actions.
The quality and quantity of training data are the single greatest determinants of ML model performance. A powerful algorithm trained on poor data will produce poor results — a principle sometimes summarized as 'garbage in, garbage out.' Feature Engineering, Cross-Validation, and Hyperparameter Tuning are the core practices that transform raw data and algorithms into reliable, deployable systems.
Machine Learning replaces hand-coded rules with data-driven pattern recognition — making it possible to automate tasks that are too complex or variable to program explicitly.