Supervised learning

Supervised learning is a foundational approach in Machine learning where an algorithm learns patterns from labeled training data—examples paired with known correct answers. The learner uses these labeled examples to discover the relationship between inputs and outputs, then applies that knowledge to make predictions on new, unseen data.

Think of it like learning with a tutor who provides immediate feedback. A model might learn to recognize Honey Bees by studying thousands of labeled bee photographs, or predict house prices by analyzing historical sales data paired with their actual costs.

The process typically involves three stages: training on labeled data, validating on a held-out set, and testing on completely fresh examples. Common techniques include Neural Networks and their deeper variants, Deep neural networks, which use mathematical optimization methods like Backpropagation to adjust internal parameters and minimize prediction errors.

Supervised learning powers countless real-world applications—from medical diagnosis and internet spam filters to AI systems that generate text. The quality and quantity of labeled data critically determines success; gathering good labels remains one of the field's practical bottlenecks.

Contrasts with Machine learning's other major paradigm, unsupervised learning, where no labels guide the discovery process.

Related

Artificial Intelligence, Deep neural networks, Backpropagation, Neural Networks, Machine learning